Looking at all detected genetic markers we could describe two major types of marker gene combinations represented by group 1A and group 2B. All other groups depict a gradual transition of marker gene combinations between these two groups. Thus the main focus on attention should be on these two groups. Group 1A is characterized by the presence of cj1365c, cj1585c, dimeric tlp7[2], cj1321- cj1326, fucP, cj0178, cfrA/cj755, and ceuE
11168
as well as the absence of ansB, dmsA, ggt and cstII. In contrast to that, group 2B is an inverted mirror image of this constellation: positive for ansB, dmsA, ggt but negative for cj1365c, cj1585c, dimeric tlp7[2], cj1321- cj1326, fucP, cj0178, cfrA/cj755, ceuE
11168
as well as cstII/III.
Champion and coworkers identified the flagellin O-glycosylation locus cj1321-cj1326 as marker present in livestock-associated strains, whereas 55.7% of clinical isolates were shown by them to be negative for this gene cluster [6]. According to their data, cj1321-cj1326-negative strains originate mostly from asymptomatic carriers and the environment [6]. Due to our data, 63.9% of the tested C. jejuni isolates show livestock association based on the presence of cj1321-cj1326. But in contrast to their findings the non-livestock-associated group 2B is significantly more often associated with human origin and thus, bears obviously higher pathogenic potential for humans than the livestock-marker positive strains.
The fucP gene was shown to be present only in isolates negative for ggt[8], which is in accordance with our findings. The ggt-positive group 2 is almost completely free of fucP-positive isolates. Interestingly, group 6 isolates, positive for the ggt-associated marker genes ansB and dmsA but not for ggt, are mostly able to utilize L-fucose. The fucP distribution pattern is similar to that of the livestock-association marker genes cj1321-cj1326 and the serine protease Cj1365 [2]. Thus, fucP should be considered as a further marker for livestock association. It can be suggested that the fucose permease is a crucial prerequisite for dwelling in the mucosa layer, while it enables the bacterial cell to metabolize mucosal L-fucose.
The ability to acquire iron is an essential prerequisite for bacterial replication and thus an important virulence factor especially in iron restricted environments [17, 18]. While C. jejuni has no own siderophores [10] it makes use of exogenous siderophores produced by accompanying bacterial species [19]. At all five different iron uptake systems have been detected in the genome of C. jejuni NCTC 11168 [10], but the genome sequence of strain 81–176 reveals three fundamental differences in this regard [9]. Cju15, a protein of unknown function, replaces the gene cfrA/cj0755, which encodes a ferric uptake receptor [9]. A second iron uptake transport system encoded by the genes cj0173c-cj0182 is missing critical components e.g. cj0178 and tonB3[9], and in the gene cluster encoding the enterochelin uptake system cju30 is inserted between cj1355 and cj1356c[9]. Additionally the enterochelin uptake system (CeuBCDE; Cj1352 to Cj1355) is ubiquitous within the C. jejuni population, but it shows sequence variability detectable by PCR using different primers. A C. jejuni subpopulation, associated with a higher rate of bloody diarrhea requiring hospitalization, was identified by Feodoroff and coworkers [7]. This subpopulation was positive for ggt, but ceuE was not detectable using ceuE-primers derived from the NCTC 11168 genome sequence. This subpopulation corresponds to group 2 in our scheme. In a significant number of group 2 isolates it was only possible to detect the ubiquitous gene for ceuE using primers derived from the genome sequence of C. jejuni strain 81–176 (for pldA we detected no significant differences). In this group of isolates the iron uptake system components cj0178 and cfrA/cj755 are absent in nearly 100% of the isolates. Thus, the two groups identified by Feodoroff et al. associated with bloody stools/GGT-production and an increased hospitalization rate/ceuE11168-presence overlap to a larger part that corresponds to group 2B. Besides ggt and ceuE11168, cj0178 and cfrA/cj755 should be considered as marker genes correlating with clinical data.
Parker et al. defined, according to the organization of the LOS locus, various LOS locus classes (LLC). The LOS locus of LLC A, B, C, M and R includes the sialic acid synthase (neuBCA) and two class-specific sialyltransferases: cstII in LLC A, B, M, R and cstIII in LLC C [19, 20]. It was demonstrated that the LOS plays a role for epithelial cell invasion [4] and is associated with the clinical course of gastro-enteritis [5]. In this study, we detected just the key-enzymes for LOS sialylization cstII and cstIII. Besides the isolates of the groups 2B and 6, the test population was either cstII or cstIII positive. Group 1A and 1B* isolates were predominantly positive for cstIII. This corresponds to the results of Habib et al. that CC 21 belongs to either LCC C or LCC A [3]. The subgroup 1B**, consisting of CC 48 and 206 isolates, is only cstII but not cstIII positive, corresponding mostly to LLC B [3, 15]. The isolates of the subgroup 1B*** (CC 49 and CC 446) were demonstrated to be partially positive, partially negative for cstII but generally cstIII-negative. This corresponds to LLC B and D due to few isolates described by Habib et al.[3]. The majority of group 2A isolates was tested positive for cstII, corresponding to LCC A1 and B [3, 16] in contrast to group 2B isolates that were tested negative for both cstII and cstIII and belong to LLC D and E(H) [3]. Positive tested for cstII but not cstIII was the majority of isolates in group 3. An exclusion were the isolates of CC 353 that are cstIII-positive (corresponding to LCC C). The negative test result for cstII- and cstIII of the majority of isolates in the groups 4, 5, and 6 implies that they belong to LLCs with non-sialylated LOS. Hotter et al. associated LCC D and E, corresponding to group 2B in our study, with an increased hospitalization rate [5], that is in accordance with the results obtained by Feodoroff and coworkers for the ggt-positive and ceuE11168-negative group [6] as well as with our prevalence rates for isolates of human origin. In contrast to our data and the data of Feodoroff et al.[7] Hotter and coworkers associated LCC B and C with a higher frequency of bloody stools [5]. This group of isolates corresponds for the most part to the group 1 but also 2A.