Comparison of Campylobacter fla-SVR genotypes isolated from humans and poultry in three European regions

Authors


Trudy M. Wassenaar, Molecular Microbiology and Genomics Consultants, Tannenstrasse 7, 55576 Zotzenheim, Germany. E-mail: trudy@mmgc.eu

Abstract

Aims:  The genetic diversity of Campylobacter isolated from human infection and from poultry was assessed in strains originating in three different European regions in order to compare these two hosts and to investigate European regional differences.

Methods and Results:  Randomly chosen isolates originated from Norway, Iceland and Basque Country in Spain were genotyped by sequencing of the short variable region (SVR) of flaA. A total of 293 strains were investigated, c. 100 per country with half originated from either host. The results indicate extensive diversity in both hosts and identified differences in the nature and distribution of genotypes between the countries. These differences could in part be related to geographical location, in that Campylobacter genotypes from Iceland and Norway were more similar to each other than either was to Basque Country.

Conclusions:  Differences between the countries exceeded the observed differences between human and poultry isolates within a country.

Significance and Impact of the Study:  Regional differences are extensive and should not be ignored when comparing genotyping data originating from different international studies.

Introduction

The genetic diversity of Campylobacter jejuni and Campylobacter coli from humans and poultry is well established. Whatever genotyping method is applied, invariably a large diversity is observed, the degree of which seems mostly determined by the discriminatory power of the method (Wassenaar and Newell 2000). In most studies, overlap of genotypes was observed for human and poultry isolates, but nearly always that overlap was incomplete so that human- and poultry-specific genotypes were also observed (see for instance, Moore et al. 2003; Siemer et al. 2005; Fang et al. 2006; Zorman et al. 2006; Keller et al. 2007; Lévesque et al. 2008).

The method of multi-locus sequence typing (MLST) is most appropriate to study population biology of bacterial subpopulations in various environments (Maiden et al. 1998; Maiden 2006) and has been developed for Campylobacter as well (Dingle et al. 2001). However, this method is elaborate and expensive and investigates the alleles of a limited number of housekeeping genes that are under negative selective pressure. Although this gives valuable information about the population structure of investigated samples, it is of limited use for short-term, localized epidemiological studies. In the latter case, a cheap and easy genotyping method determining alleles of virulence or colonization factors would be appropriate. Determination of the sequence of a short variable region (SVR) in the gene encoding flagellin A, called fla-SVR, fits these requirements (Meinersmann et al. 1997).

In this study, we compare and contrast Campylobacter isolates by fla-SVR genotyping, obtained from the human and chicken host in three European regions: Basque Country in Spain, Iceland and Norway. In all three regions, most of the consumed poultry is produced nationally. The samples were taken during 2002–2006 and samples were selected randomly from human cases of diarrhoea that were bacteriologically investigated and from local poultry or poultry products. The setup of this work was designed neither to draw conclusions on the likelihood that chicken consumption can result in human disease nor to identify outbreaks or to assess source attribution. Instead, the aim was to compare the genetic diversity in these two hosts in three different European regions, to compare these two hosts and to investigate European regional differences.

Methods

The origin of 293 Camp. jejuni isolates included in this study is summarized in Table 1. It was not recorded whether the strains of human origin, all from clinical cases, were domestically acquired, or whether the cases were related to chicken consumption. Ten randomly selected samples of uncomplicated, bacteriologically confirmed campylobacteriosis were taken per year for the years 2002–2006.

Table 1.   The origin of 293 Campylobacter isolates included in this study, all isolates originated during 2002–2006
Country of originIsolation sourceNo. of isolates
Basque CountryPoultry50
Basque CountryHuman50
NorwayPoultry48
NorwayHuman49
IcelandPoultry47
IcelandHuman49

The poultry isolates from Basque Country were obtained from whole carcass, sausage, wings, thighs and hamburgers (containing chicken meat only) at retail level from local markets. The poultry isolates from Norway were taken at slaughter as cloacal swabs in the years 2002 and 2003 and caecal samples in 2004–2006. All poultry isolates from Iceland were caecal samples taken at slaughter.

Genotyping was performed by PCR amplification and sequencing of the SVR of flaA (Meinersmann et al. 1997) and the nomenclature of the Oxford database (http://pubmlst.org/campylobacter/flaA) was used. Peptide sequences were translated from the obtained DNA sequences and named according to the Oxford database.

Statistical analysis was carried out using the function fisher test in R ver. 2·7·1 (R Development Core Team, 2008). The comparison was drawn between numerical values of observations of one genotype or peptide compared to the whole dataset, for each country separately, or for each host separately, as indicated. All differences reported as ‘significant’ produced a P < 0·05.

Presence of the genes virB11 (Gilbert et al. 2000), cgtB and wlaN (Müller et al. 2007) was determined by PCR and presence of a PCR product was recorded in a binary manner, with positive when a product was present or negative when absent for each reaction. These binary results were then compared for isolates sharing an identical fla-SVR genotype; when the three results were identical, this produced a mismatch score of zero, but when a difference in one of these results was detected, it was scored as a single mismatch.

Results

Genotypes of all isolates combined and analysed per region

The distribution of genotypes for 293 samples originating from Basque Country, Norway and Iceland is shown in Fig. 1. A total of 92 different alleles were detected, of which 51 genotypes were present only once. The most frequently detected genotypes were genotype 36 (gt36) with 25 isolates or 8·4%, gt32 (20% or 6·7%), gt34 (19% or 6·3%), gt15 (14% or 4·7%), gt239 (13% or 4·3%) and gt8 (12% or 4·0%). These six genotypes, together covering 34·4% of all isolates, were not always common in the three individual regions, as can be seen from their distribution per region shown in Fig. 2. Regional differences in the distribution were statistically significant (P < 0·05) for genotype 32 that was found overrepresented in Norway, for gt34 that was exclusively detected in Basque, and for gt8 and gt15 that were found absent and uncommon in Basque respectively. The other common genotypes gt36 and gt239 did not significantly differ between the three regions.

Figure 1.

 Distribution of 92 allelic genotypes in 293 isolates of Campylobacter jejuni. Each colour represents a genotype (gt) as indicated in the legend, which also lists how often these were found (number). Twelve genotypes that were present twice and 51 isolates found only once are represented as combined segments.

Figure 2.

 The 34% most frequent genotypes, by country. The distribution of gt32, 34, 15 and 8 is significantly (P < 0·05) different between countries. (Grey) Iceland; (black) Norway and (white) Basque Country.

The distribution of all genotypes detected per region is represented in Fig. 3. The variation in common and less common genotypes is quite extensive, and regional differences are substantial in the genotype distribution of Campylobacter. In Basque Country, a total of 38 different genotypes were detected, of which 23 were represented in single isolates. In Norway, 44 different genotypes were identified including 27 singles, whereas in Iceland 29 of 49 detected genotypes were singles. The absence of gt34 from Iceland and Norway, and the absence of gt8 from Basque Country are statistically significant observations.

Figure 3.

 Distribution of genotypes in the three regions investigated. The colour codes of genotypes that were detected more than once are standardized between the three charts (and the chart in Fig. 1) and their percentages are given in parentheses for each region.

Analysis of genotype per host

Genotypes were analysed with respect to the host from which the isolate had been derived. In total, 53 different genotypes were observed from human cases irrespective of country of origin, of which 30 were detected only once (results not shown). The most common genotypes were 32, 34, 36, 15, 92 and 239, summing up to 40·9% of the human isolates. In poultry, 67 different genotypes were observed with 41 singles. The most common genotypes in poultry were 36, 8, 70, 21, 32 and 34, which in combination represented 32·3% of the poultry isolates. Thus, the six most common genotypes in humans covered a larger fraction of isolates in that host than the six most common genotypes in chickens, and in total there were more genotypes detected in poultry than in humans. Although these data suggest that the genetic diversity of Campylobacter in poultry exceeds that of humans, regional differences should be taken into account when comparing the two hosts.

Comparing the diversity of Campylobacter isolates per country and host, it is observed that in Basque Country indeed the total number of genotypes detected in poultry is bigger than in humans, but this observation is not significant. In Iceland and Norway, isolates from the two hosts display the same degree of variation (Table 2). The contribution of either host for the most common genotype per region (covering at least 50% of all detected genotypes in that region) is shown in Fig. 4. For Basque Country, six genotypes cover 50% of the isolates and all of these are found in both hosts. Fifty per cent of the isolates from Norway are covered by the nine most common genotypes, again detected in both hosts. In Iceland, 55% of the regional isolates are divided over the 12 most common genotypes that all have relatively few numbers.

Table 2.   Genetic variation in Campylobacter isolates from human and poultry per country
CountryBasque CountryNorwayIcelandIn all three countries
Total number of genotypes in human isolates20273053
Total number of genotypes in poultry isolates29302967
Total number of genotypes38444992
Figure 4.

 The most common fla-SVR genotypes that together represent at least 50% per country are divided in human and poultry isolates. The scale is kept constant for all three diagrams. (Light grey) Human isolates and (dark grey) poultry isolates.

Analysis by year and by genotype

The most common genotypes were observed throughout the investigated period although some genotypes peaked in a particular year in a particular region. In order to investigate whether identical sweeps of genotypes were shared between regions, the genotypes identified in all 5 years in each region were plotted (Fig. 5). Conserved trends could not be recognized for these common genotypes in these three regions. This temporal analysis again emphasizes local differences in genotype distribution between the analysed countries.

Figure 5.

 Temporal distribution of genotypes that were found in five consecutive years in the three regions irrespective of the host. (Grey) Iceland; (black) Norway and (white) Basque Country.

Analysis of peptide type coded by alleles

Because flaA codes for the immunogenic protein flagellin, we also considered the peptide that was coded by the various alleles (Table 3). Several genotypes are now combined in a single peptide type because of redundancy of the genetic code. In total, 40 different peptides were recognized, of which 28 were found in poultry and 27 in human isolates. By far, the most common peptide and statistically highly significant was peptide 1 (P = 5·6 × 10−9), which was evenly distributed over human and poultry isolates as well as between countries (Table 3). Two peptide types displayed a significant host-specific distribution: peptide 27 was only observed in poultry and peptide 11 was only observed in human. Five peptides displayed regional distribution differences: peptide 12 was only observed in Basque, peptides 18, 15 and 118 were only observed in Norway and Iceland, and peptide 8 was only observed in Iceland and Basque Country. Although all of these observations were significant, there is currently no explanation for such regional differences.

Table 3.   Distribution of most frequently observed peptide sequences, representing 88% of all data
Peptide type (coded by No. of alleles)Total No. identifiedNo. in Basque CountryNo. in IcelandNo. in Norway Percentage in poultry*
  1. Statistically significant findings are indicated in bold.

  2. *Percentage of peptides in poultry isolates is given irrespective of country.

1 (18)11041323751
5 (12)4714181553
9 (3)1554640
2 (1)1354469
8 (4)1082060
18 (1)1007360
10 (6)962167
3 (4)924322
11 (16)73400
15 (5)706143
24 (1)602417
27 (1)5122100
118 (1)5014100
12 (1)550040
All peptides combined:40 different peptides17 different peptides23 different peptides26 different peptides28 in poultry, 27 in human isolates

Validation of isolate clustering by fla-SVR typing

The fla-SVR genotyping method zooms in at a single genetic allele, whereas it is known that Campylobacter DNA can be subjected to recombination as well as mutations (Wilson et al. 2009). In case the flaA gene would undergo such genetic events, incorrect grouping could result that no longer reflect the true genetic similarity of the isolates. Recombinations and mutations become more frequent when investigating samples originating from various locations during a long time span, as is the case in this investigation. In order to assess how reliable fla-SVR grouping is under these conditions, all Basque and Norwegian isolates were subjected to three alternative genotyping methods (the isolates from Iceland were not included in this analysis). Presence of virulence genes virB11, cgtB and wlaN was determined by PCR and presence of a PCR product was scored as positive. These three virulence-associated genes were selected as other studies have shown that they are frequently absent in Campylobacter (Kordinas et al. 2005; Datta et al. 2003; Müller et al. 2006; Talukder et al. 2008). Hence, absence of a PCR product would most likely be caused by absence of the gene, although it was not investigated whether primer mismatches were the basis of negative PCR results. The Basque isolates were then grouped for identical fla-SVR allele, and for any fla-SVR allele found more than once, the number of mismatches found in the other three genotyping tests were scored as described in Methods. In this way, 77 Basque strains were compared (23 strains with a single fla-SVR alleles could not be assessed) and eight mismatches were identified. This was interpreted to mean that fla-SVR grouping could be inadequate in at least 8 out of 77 or 10% of cases: in these instances, we had evidence that the isolates were not genetically identical although they shared the same fla-SVR. As the PCR tests of the three genes were frequently negative (68 of 100 strains were negative for all three genes), the correlation between combined virB11, cgtB and wlaN genotype and fla-SVR could not be assessed in the reverse direction. The same analysis of the 69 Norwegian strains that shared a fla-SVR type produced 20 mismatches in the combined virulence gene test, or 29%. This weaker performance would suggest that the fla-SVR genotype is a less reliable marker for the Norwegian strains than for the Basque strains. Possibly, recombinations or mutations have occurred more frequently in the investigated Norwegian Campylobacter population than in the investigated Basque population.

Discussion

Multiple studies have demonstrated the genetic diversity of Campylobacter causing human infections, present in live poultry and on poultry meat. Few analyses, however, encompassed analysis of isolates from various countries in a standardized and systematic way so that little information is available about diversity between countries. The use of standardized genotyping methods, especially those based on DNA sequences that allow objective interpretation, and standardized nomenclature such as provided by MLST and fla-SVR genotyping, now allows comparison of genotyping data between countries.

This study compared Campylobacter by fla-SVR genotyping originating from human and poultry sources in three European regions where epidemiological differences were expected. Iceland provides a relatively closed environment with severe winters during which Campylobacter is infrequently found in local poultry. The strong summer peak in Icelandic human cases is most probably related to travel (Anonymous, 2008); Norway shares a northern climate with Iceland, whereas Basque Country has a southern climate with mild winters during which campylobacteriosis is still relatively common (Anonymous, 2007). From each region, ten isolates were isolated per year per source, whereby seasons yielding more isolates were represented more frequently in the selection. Nevertheless, the method of isolation and the poultry sources differed per country, as the study was designed retroactively. Therefore, interpretation of the findings needs to be carried out with care, as the investigated isolates from human infections, food sources and birds at slaughter may have undergone potentially different selective bottlenecks. For instance, all investigated human isolates caused disease, so this may reflect a selection for virulence, whereas isolates on foods have all been exposed to cooling of the product, and caecal isolates were all effective chicken colonizers.

From our analysis, we conclude that regional differences exist in the population diversity of Campylobacter, but common genotypes (assessed by a single gene allele) were also found present in the three investigated countries. Regional differences in part correlated to geographical location and may be related to commonness in climate. For instance, Iceland and Norway resembled each other more closely than either country resembled Basque Country, in terms of genetic diversity of the Campylobacter genotypes encountered, but not in terms of shared genotypes. Despite the relative isolation of Iceland, the number of different allele types detected in either the poultry or human host is not significantly less than that encountered in the other two countries. The total number of genotypes identified in Iceland was higher than in the other countries despite similar numbers in poultry, presumably because of a relative large number of Icelandic travel-associated human cases.

Few significant differences in the distribution of Campylobacter genotypes over the two hosts were observed, and regional differences have to be taken into account for such an analysis. A correlation between the flagellin peptide sequence and the host was identified for two peptide types, although the most common peptide was evenly distributed between human and poultry. Because the fraction of flaA sequenced in fla-SVR is located outside recognized immunodominant regions of the protein (Fernando et al. 2008), it is unclear whether observed host differences are driven by immunological selection.

The results of single-locus fla-SVR typing were validated by amplification of three further genetic markers, which identified potentially incorrect strain identification in at least 10% of Basque cases and in at least 29% in Norwegian cases. Thus, even in this respect local differences can be observed. Similar observations were described by Merchant-Patel et al. (2008) who tested the presence of single nucleotide polymorphisms in isolates with identical flaA RFLP (banding pattern) genotypes and found these to differ in epidemiologically unrelated isolates.

The observations presented here have major implications on the interpretation of molecular epidemiology of Campylobacter in Europe, as it implies that findings generated in one country cannot be taken to exemplify the situation in another country. Comparison of genotypes between countries or even continents must be carefully interpreted (McTavish et al. 2008). Even within one country or region, temporary variation may be extensive. Although this conclusion is based on a single genetic marker, in this case fla-SVR, it can be argued that the same could apply for a multi-locus typing method, such as MLST, as we believe the results are mostly reflective of true differences between the regional Campylobacter populations in the three regions of investigation.

Acknowledgements

This study was performed with support from CampEC-NET, a project within the SAFEFOODERA network.

The following colleagues are acknowledged for providing strains: for Spain: J. Cárcamo (Normative Laboratory of Public Health, San Sebastian), A. Canut (Hospital Santiago, Vitoria-Gasteiz), and L. Michaus (Hospital Txagorritxu, Vitoria-Gasteiz); for Iceland: E. Gunnarsson (Institute of Experimental Pathology, Keldur) and H. Hardardottir (Institute of Laboratory Medicine, Department of Clinical Microbiology, Landspitali University Hospital); for Norway: G. Kapperud (The Norwegian Institute of Public Health) and B. Bergsjø (National Veterinary Institute).

Ancillary