Lipooligosaccharide locus classes and putative virulence genes among chicken and human Campylobacter jejuni isolates

Campylobacter cause morbidity and considerable economic loss due to hospitalization and post infectious sequelae such as reactive arthritis, Guillain Barré- and Miller Fischer syndromes. Such sequelae have been linked to C. jejuni harboring sialic acid structures in their lipooligosaccharide (LOS) layer of the cell wall. Poultry is an important source of human Campylobacter infections but little is known about the prevalence of sialylated C. jejuni isolates and the extent of transmission of such isolates to humans. Genotypes of C. jejuni isolates from enteritis patients were compared with those of broiler chicken with pulsed-field gel electrophoresis (PFGE), to study the patterns of LOS biosynthesis genes and other virulence associated genes and to what extent these occur among Campylobacter genotypes found both in humans and chickens. Chicken and human isolates generally had similar distributions of the putative virulence genes and LOS locus classes studied. However, there were significant differences regarding LOS locus class of PFGE types that were overlapping between chicken and human isolates and those that were distinct to each source. The study highlights the prevalence of virulence associated genes among Campylobacter isolates from humans and chickens and suggests possible patterns of transmission between the two species.


Background
Campylobacter is the most common cause of bacterial enteritis in the western world accounting for more than 200 000 reported human infections per year in the EU [1]. Campylobacter jejuni and C. coli are the two most common species infecting humans. The disease is normally self-limiting and resolves within 2-3 weeks. However, in some cases patients need to be hospitalized due to severe enteritis or to sequelae such as reactive arthritis, Guillain Barré-or Miller Fischer syndromes [2][3][4][5]. The total economic burden of campylobacteriosis has been estimated to appr. 2.4 billion euros annually in the EU [6]. Campylobacter is a zoonotic pathogen and the sources of infection mainly include products from farm animals and contaminated water. In many countries, poultry is the single most important source with broiler meat estimated to account for 20-30 % of human cases in the EU [6]. The prevalence of Campylobacter generally shows a strong seasonal variation in European countries, with a peak between June and September, both in terms of reported human cases and positive chicken flocks [1]. The mechanisms behind the pathogenicity of C. jejuni are poorly understood. The existence of virulence associated genes has been proposed in several studies but few have been correlated to disease severity in clinical materials [2,4,5,[7][8][9][10]. The structure of the lipooligosaccharide (LOS) layer of C. jejuni has been suggested to be an important determinant of the pathogenicity of certain postinfectious sequelae [4,5,10]. The synthesis of LOS in the outer membrane of C. jejuni is regulated by the LOS operon and based on the genomic organization, this operon has been categorized into 19 different LOS locus classes denoted A to S [11,12].
Strains belonging to LOS locus class A, B and C harbor genes that enable sialylation of the LOS structures. Such strains have been associated with the post-infectious sequelae Guillain Barré-and Miller Fischer syndromes as well as with reactive arthritis [4,5,10]. Some studies have also proposed that C. jejuni strains with sialylated LOS might be involved in more severe symptoms of enteritis as well as invasive infection [5,13,14]. However in previous studies, we have not been able to find any associations between such strains and severe enteritis or with invasive infection in patients [7,8]. The prevalence of C. jejuni strains with genes for sialylation of LOS among enteritis patients seems to vary between 50 and 60 % [5,8,15]. Although poultry is an important source of Campylobacter infection in many countries, very little is known about the distribution of LOS locus classes among C. jejuni from broiler flocks reared for consumption [15,16]. In the present study, we compared isolates from Swedish enteritis patients with domestically acquired C. jejuni infection to C. jejuni isolates from broiler chickens raised in Sweden, with the aim to study patterns of virulence associated genes and their possible involvement in transmission from chickens to humans in Sweden.

Species identification and PFGE typing
A total of 163 C. jejuni isolates were included in the study, including 104 from chicken and 59 from humans. Pulsed field gel electrophoresis (PFGE) showed that the 163 C. jejuni isolates fell into 66 different SmaI types. Of these, 19 types were unique for humans (24 isolates) and 32 types were unique for chicken (43 isolates). Fifteen SmaI types were detected among isolates of both human (32 isolates) and chicken (57 isolates) origin. The unique SmaI types each contained from 1 up to 3 human isolates or from 1 up to 4 chicken isolates. The shared SmaI types contained from 1 up to 7 human isolates and from 1 up to 19 chicken isolates. Seven isolates (3 from humans and 4 from chickens) were not cleaved by SmaI and hence, the SmaI type could not be determined. In total, 55 % of all isolates belonged to SmaI types that were shared between human and chicken isolates.

Lipooligosaccharide (LOS) locus classes
All isolates were screened for LOS locus class by PCR. In general the distribution of LOS classes were similar among the human and chicken isolates. LOS locus class C was the most common in both materials followed by class H. The LOS locus classes with genes for incorporation of sialic acids in the LOS (classes A, B and C) together constituted 57 and 50 % of the human and chicken isolates respectively (Table 1). Temporal variations in the occurrence of the different LOS classes were analyzed but no apparent seasonal differences were observed among either human or chicken isolates.
As expected, there was generally a strong coherence between LOS locus class and SmaI type where the vast majority of the isolates in one particular SmaI type had the same LOS locus class (data not shown). To study if certain LOS locus classes were connected to the presumed chicken source of the human isolates in this material, we analyzed the association of LOS locus class with isolates that were shared between human and chicken isolates and those that were unique to the respective source in this material. All SmaI types with more than 2 isolates, in total 26 (11 unique and 15 shared) were included in the comparison, covering altogether 123 isolates (44 human and 79 chicken). Among isolates with LOS locus class C there were no significant differences in the proportions belonging to the shared and the unique SmaI types (P = 0.120, Table 2). Isolates with LOS locus class H were significantly more common among SmaI types that were shared among chicken and humans than among SmaI types that were unique to each species (P = 0.014). LOS   [11] and an analysis of these two classes as one group showed a highly significant overrepresentation of such isolates among the shared SmaI types (P = 0.001). On the other hand, isolates with LOS locus class B were more common among SmaI types that were unique for chickens and humans, as compared to shared SmaI types (P = 0.0009). This was true also for the ORF18 positive isolates (P < 0.0001).

Putative virulence genes
The prevalences of the putative virulence genes were generally similar between the human and chicken isolates ( Table 3). The gene fucP was detected in 64 % of the human-and 55 % of the chicken isolates respectively (P = 0.232), whereas virB11 was not found at all among the human isolates and only in 11 % of the tested chicken isolates. The genes ceuE, pldA and ciaB were present in the vast majority of both human and chicken isolates. We analyzed if the presence of these genes in the C. jejuni isolates was related to season but no significant seasonal variation was found. The largest difference was found between fucP positive vs fucP negative among human isolates in June (P = 0.281). As for the LOS locus classes, there was strong coherence between the putative virulence genes and SmaI type (data not shown). Furthermore, the presence of the gene fucP was strongly associated with certain LOS locus classes, both in the human and chicken isolates (Table 4). In particular, isolates of LOS locus class C were almost exclusively fucP positive, whereas nearly all isolates of LOS locus classes E and H were fucP negative. Isolates with LOS locus classes A and B were generally fucP positive.

Discussion
In this study on human and chicken C. jejuni isolates obtained during year 2009 and analyzed by PFGE, we found that the majority of isolates (55 %) had SmaI types that were shared between human and chicken isolates.
Assuming that this is a result of transmission between the two species, it supports the current notion that chicken is a major source of human C. jejuni infections. Interestingly, we found that the number of SmaI types that were unique either to chickens (n = 32) or to humans (n = 19) were higher than the number of SmaI types that were shared between chickens and humans (n = 15). This could suggest that there are only certain genotypes among the Swedish chicken isolates that are more likely to be transmitted to humans. When studying the distribution of LOS locus classes as well as the presence of the putative virulence genes fucP, ceuE, virB11, pldA and ciaB in the human and chicken isolates it was found that the LOS locus classes showed similar distributions among the human and chicken isolates. The LOS locus class C was the most common in both sources followed by the classes E and H, which is in line with the results from other studies on human isolates [5,8,15]. By a combined analysis of LOS locus classes and the SmaI restriction patterns, we sought to identify potential differences in transmission of isolates with different LOS locus classes between the Swedish chickens and humans as well as from unknown sources. As the chicken isolates in this study represented approximately 99 % of all chickens that were produced in Sweden during 2009, and the human isolates represent all patients in a specific region that have sought medical care for gastrointestinal symptoms attributed to domestically acquired C. jejuni infection, it could be assumed that a human C. jejuni isolate with a SmaI pattern identical to a chicken isolate most likely actually originated from this chicken source. Human or chicken isolates with SmaI patterns that were unique to either source on the other hand, could be assumed to originate from other sources including environmental, other farm animals or imported chicken. With this analysis we found that isolates of LOS locus class C were equally represented among isolates with SmaI types that were shared between the Swedish chickens and humans as among those that were unique, suggesting that such isolates are readily transmitted between the Swedish chickens and humans, but were also acquired from other sources. Isolates of LOS locus class E and H were significantly more likely to have shared SmaI patterns than  Isolates with LOS locus class B on the other hand showed an inversed distribution with significantly more isolates having SmaI patterns unique to either chickens or humans, suggesting that such isolates were rarely spread between the two species in this material. The same was true for the ORF 18 positive isolates although this group does not represent a single LOS locus class.
In line with the results from previous studies, we found that nearly all isolates, both from humans and chickens had the gene ceuE, suggesting that this gene is important both for human infection and colonization of chickens. The ability of C. jejuni isolates to utilize Lfucose, mediated in part by the fucose permease gene fucP, has been suggested to provide an advantage in the colonization of chickens and piglets [17,18]. In this material however, the prevalence of fucP was only slightly higher than 50 % among the human and chicken isolates. In our previous studies we found this particular gene to be present in 49 % of isolates from enteritis patients and in 23 % of isolates from bacteremia patients in Finland [2,8]. Together these results suggest that Lfucose utilization by fucP hardly is a prerequisite for colonization of chickens or for causing infections in humans.

Conclusions
In conclusion, we found that the prevalences of the different LOS locus classes in general, and of the putative virulence factors fucP, ceuE, pldA and ciaB were similar among human and chicken C. jejuni isolates in Sweden. However, LOS locus classes E and H were more abundant among isolates with SmaI types that were shared between Swedish chickens and humans whereas LOS locus class B was more abundant among isolates with SmaI types unique to the sources. Altogether, these results suggest that humans might acquire C. jejuni isolates with different LOS locus classes from different sources. Taking into account the association between C. jejuni isolates with some LOS locus classes and post infectious sequelae such as Guillain Barré syndrome and reactive arthritis in humans, this is an interesting finding that needs further studies in larger materials.

Bacterial isolates
Human isolates of domestic origin (as determined from the travel history of the patients) during 2009 (N = 60) were obtained at the Clinical Microbiology Laboratory at Uppsala University Hospital. Chicken isolates were collected within the frame of the Swedish Campylobacter Monitoring Programme for broilers. This has been run by the Swedish Poultry Meat Association since 1991 [19]. In the programme, which includes 99 % of broilers in Sweden, samples were taken from every broiler flock at slaughter. Ten caecal samples were pooled into one composite sample from each flock. In 2009, more than 3000 chicken flocks were sampled and approximately 12 % were positive for Campylobacter. One isolate from every second positive chicken flock was selected for the study, making up a total of 110 isolates from 52 different broiler producers. Preliminary species identification was performed by microscopic examination of morphology and motility as well as oxidase, catalase and hippurate test [20,21].
All isolates were then identified to species level by PCR [22][23][24]. For this, DNA was extracted using the Qiagen blood and tissue kit (Qiagen, Sollentuna, Sweden) according to the manufacturer´s instructions. The following primer pairs were used: MDmapA1 upper-and MDmapA2 lower primers targeting the mapA gene of C. jejuni (annealing temperature (AT) =59°C) [22] as well as the COL3 upper-and MDCOL2 lower primers targeting the ceuE gene of C. coli (AT = 59°C) [22,23]. Primer pair: C412F and C1228R (AT = 58°C) targeting a region of the 16S rRNA gene specific for the genus Campylobacter was used to confirm successful DNA extraction [24]. All primers were purchased from Eurogentec (Ougrée, Belgium). Cycling conditions were 10 min at 95°C, 25 cycles of 30s at 95°C, 90s at annealing temp. and 1 min at 72°C, with a final extension step of 10 min at 72°C. The strains C. jejuni NCTC11168 (ATCC 700819), C. jejuni 81176 (ATCC BAA-2151) and C. coli LMG6440 were used as positive controls and omission of template was used as negative control. Seven isolates (1 human and 6 chicken) were determined as C. coli and excluded from the study.

Genotyping of C. jejuni by Pulsed field Gel Electrophoresis (PFGE)
All isolates were analyzed by PFGE according to standard protocols. The restriction enzyme SmaI was used for cleavage of whole genomic DNA obtained from C. jejuni cultured on blood agar plates. For electrophoresis the instrument CHEF-DR II (BioRad, Stockholm, Sweden) was used. The BioNumerics program version 7.5 (Applied Maths, Kortrijk, Belgium) was used for analysis of restriction patterns of the isolates. Isolates with identical restriction patterns were designated as belonging to the same SmaI type.

Putative virulence associated genes
All C. jejuni isolates were analyzed by PCR for the presence of genes encoding the putative virulence factors, fucose permease (fucP) and the iron transport protein (ceuE). Furthermore all human isolates and 45 chicken isolates were screened for presence of the genes phospholipase (pldA), the Campylobacter invasion antigen (ciaB) and the plasmid borne virulence gene (virB11). A close to 100 % prevalence of the genes pldA and ciaB as well as very low prevalence of virB11 were expected among the chicken isolates from previous studies, and hence only a subset was selected to verify this [25]. The reaction mixture was 1x AmpliTaq Gold 360 buffer with 1.25U of AmpliTaq Gold 360 polymerase (Applied Biosystems, Austin, USA), 200 μM dNTP (Fermentas, St. Leon-Rot, Germany), 0.2 μM of each primer (Eurogentec, Ougrée, Belgium) and 5 μl of template DNA in a total volume of 25 μl. Cycling conditions were 95°C for 10 min followed by 25 cycles of 95°C for 30s, annealing temperatures (fucP: 58°C, ceuE: 60°C, pldA: 45°C, ciaB: 58°C, virB11: 53°C) for 30s and 72°C for 60s. For virB11 a touch down protocol was run with 5 cycles at 53°C, 5 cycles at 52°C and 15 cycles at 51°C. All reactions ended with an extension step at 72°C for 7 min. C. jejuni NCTC 11168 and C. jejuni 81176 were used as positive controls and omission of template was used as negative control.

Determination of LOS locus class
LOS locus class was determined for all isolates by PCR screening as described by Parker et al. [11,12]. Primer names and sequences were as described in [12]. Amplification of the open reading frame (orf ) 12 (waaV) was used to verify successful DNA extraction. The LOS locus classes A and B were identified by the presence of orf7ab (cstII), orf6ab1 (cgtB-1), orf6ab2 (cgtB-2) and orf5bII (cgtA2). Class C was identified by the presence of orf6c (cgtB) and orf7c (cstIII). The classes E, H, O and P were identified by the presence of orf26e and orf27e. The orf18df, present in isolates of the nonsialylated classes D, F, I, J, K, N, S, or Q was used for isolates that were untypeable with the primers above. Cycling conditions were 95°C for 10 min followed by 30 cycles of 30s at 94°C, 30s at 52°C, and 60s at 72°C and final extension at 72°C for 7 min. Primers for orf6ab1 and orf6ab2 were run together in a duplex PCR with an annealing temperature of 59°C.

Statistical analysis
Statistical analyses were performed with GraphPad Prism version 5 (GraphPad Software, San Diego, CA, USA) The χ 2 or Fisher's exact tests were used for comparison of categorical data. A P value < 0.05 was considered significant.