INTRODUCTION

Top Abstract Introduction Materials & Methods Results Discussion References

The gram-negative spirillum Campylobacter jejuni is a primary etiologic agent of human gastroenteritis. Most campylobacter infections appear to be sporadic rather than outbreak associated, and in a majority of cases, the original source of infection cannot be determined (10). Case-control studies (4, 17) have identified several risk factors associated with campylobacteriosis, including consumption of untreated water or milk, foreign travel, and the handling or consumption of raw or undercooked poultry. Other significant but unrecognized sources of infection may also exist.

A better understanding of the epidemiology of sporadic campylobacter infection is essential for the development of intervention measures to reduce human exposure to this pathogen. The subtyping of campylobacter isolates has been valuable in the epidemiological investigation of point source campylobacter outbreaks (15, 31) and provides a means for monitoring the transmission of C. jejuni from environmental reservoirs and animal hosts into the human food chain (27). Numerous phenotyping methods for C. jejuni have been described, all of which, to some degree, fulfill the requirements of stability, reproducibility, and discrimination necessary for epidemiological analysis. However, inconsistent strain associations occur when different phenotypic techniques are used to examine identical sets of isolates (30). The development of campylobacter genotyping methods has not substantially resolved these anomalies (1, 9, 28), although some uniformity between genetic and phenotypic analyses for the most common O serogroups of C. jejuni has been reported (14, 18, 29). Progress towards methodological standardization (39) and increased awareness of the mechanisms responsible for variation in subtyping characteristics (12) may improve the quality of epidemiological information derived from the use of multiple campylobacter typing schemes.

In this report, we describe serovars of C. jejuni identified by a comparison of heat-stable (O) and heat-labile (HL) serogrouping results from over 9,000 sporadic human campylobacter isolates. An integrated analysis of expressed features (O and HL serogroups, phage groups, and biotypes) and restriction fragment length polymorphism (RFLP) genotypes was used to produce polyphasic strain profiles (41) for 52 strains from three common O/HL serovars of C. jejuni. The results of this comparative study revealed a number of novel features relevant to the epidemiology and population biology of this important enteropathogen.

	MATERIALS AND METHODS

Top Abstract Introduction Materials & Methods Results Discussion References

Bacterial strains and culture media. Campylobacter strains for serogrouping were submitted from 23 United Kingdom (UK) laboratories over the period of 1988 to 1992. All strains were random, sporadic human fecal isolates submitted as part of a UK passive campylobacter surveillance program. Serogroup and RFLP data were also included in this study from 124 systemic C. jejuni isolates (19) and 14 human isolates and 1 bovine isolate from C. Nicol (Enteric Reference Laboratory, Porirua, New Zealand). O and HL serogroup reference cultures were provided by J. L. Penner (33 strains) and H. Lior (12 strains), including C. jejuni NCTC 11168, which was the HL 4 strain.

Strain characterization. Serogrouping of 9,024 sporadic human isolates of C. jejuni was carried out at Manchester Public Health Laboratory over the period of 1988 to 1992 by using both the Penner (O) and Lior (HL) systems in accordance with standard methods (22, 33, 34). The panel of available typing sera tested for 33 O antigens associated with C. jejuni and a restricted set of 12 of the most common HL serogroups (HL types 1, 2, 4 to 9, 11, 17, 20, and 21). Frequent cross-reactions occurred between O antigens 4, 13, 16, 43, and 50, and strains agglutinating with any of these five antisera were classified within the O 4 complex. The program DataEase (DataEase International, Inc.) was used for the comparison of O and HL serogroup determinants, and multiple isolates from known outbreaks were excluded from the analysis. Significant associations (P < 0.001) between specific O and HL antigens were calculated by chi-square analysis. Identification to the species level, biotyping, and phage typing were performed at the Preston Public Health Laboratory by previously described methods (8, 35). The methods used for nucleic acid extraction, restriction endonuclease digestion, and RFLP analysis of C. jejuni have been published elsewhere (18).

	RESULTS

Top Abstract Introduction Materials & Methods Results Discussion References

Serovars of C. jejuni are characterized by associations between specific O and HL antigens. Typing results for both the O and HL serogroup antigens were obtained for 4,294 (47.6%) of the 9,024 isolates tested. Thirteen of the 33 available O antisera, including the five cross-reacting epitopes of the O 4 complex, grouped 3,845 (89.5%) of these strains and 6,222 (69.0%) of the total number of isolates. The remaining 20 O antisera grouped 449 (10.5%) of the strains with both O and HL results, and 1,454 (16.1%) of the total number of isolates (data not shown). Three hundred forty-three isolates (3.8%) were typeable by the HL system but not by the O system, while a much larger proportion (3,382; 37.5%) were typed by O but not HL antisera. There were 999 strains (11.0%) which could not be typed by either scheme.

Strains within three serovars of C. jejuni have many features in common. The polyphasic strain characteristic (Preston biotype, phage group, 16S ribotype, and E3CJC2 RFLP type) of 52 random isolates of C. jejuni belonging to three serovars, O 2/HL 4, O 50/HL 7, and O 23/HL 5, are listed in Tables 2 to 4. These strains originated from different geographical areas within the UK and abroad and were isolated from a variety of sources over the period of 1980 to 1993. 

Atypical variation of both expressed and genotypic characteristics occurs within serovars. The cross-reacting O 23 and O 36 antigens are almost identical serologically and biochemically (7), and it was the latter serotype strain which showed the serovar characteristics. The O 23 serogroup reference strain and isolate B/1916 (Table 4) were clearly unrelated to other strains of this serovar. Although both possessed the O 23 antigen, the HL antigens of these strains were atypical, demonstrating the low predictive value of inferring interstrain relationships from O serogroup data alone. However, even within true O/HL serovars, atypical strain variation occurred. Two serovar O 2/HL 4 strains, P/2255 and P/3546 (Table 2), and two serovar O 50/HL 7 strains, P/3758 and P/2587 (Table 3), possessed E3CJC2/16S RFLP types, phage groups, and biotypes that were mostly dissimilar from those of the other strains of these serovars. More limited variation was seen with isolates C/13383 and C/19584 (Table 3), which shared the distinctive features of serovar O 50/HL 7 strains, except that the expected HL 7 antigen had been replaced with HL 1 and HL 20 epitopes, respectively.

	DISCUSSION

Top Abstract Introduction Materials & Methods Results Discussion References

A number of different phenotypic approaches to strain identification have been employed in attempts to resolve the epidemiology of sporadic campylobacter infection. The two most common systems involve serogrouping based on either O lipopolysaccharide or HL surface-associated antigens. In this report, we show that the O and HL epitopes exhibit strong nonrandom associations and that many UK strains can be allocated to specific serovars. A recent report has also demonstrated O/HL serogroup associations for sporadic campylobacter isolates from the United States (32), where the patterns of serovar occurrence are broadly similar to those in the UK. These data suggest that common prevalent strains may cause a majority of campylobacter infections in both of these developed nations. The similarities between certain North American serogroup reference strains and sporadic UK isolates (18) (Tables 2 to 4) provide further support for this hypothesis.

Evidence of clonal relationships within three serovars indicates that linkage between the O and HL epitopes may represent a practical epidemiological marker. It is not known why specific O and HL antigens occur in stable, preferred associations, nor why only a small number of serovars account for a majority of isolates from human infections. The possession of specific surface-expressed serogroup characteristics may confer a general fitness phenotype, enabling colonization of a wide range of hosts. Alternatively, certain antigenic combinations may act as particularly effective adhesins, enhance invasiveness, affect complement binding and serum resistance, or otherwise function as virulence factors.

The three serovars analyzed for polyphasic markers show conservation of phage group and, to a lesser extent, biotype characteristics. The coincidence of specific phage groups within serovars may be the result of linkage, with phage adsorption characteristics directly influenced by the epitope configuration of either the lipopolysaccharide or flagellar antigens. Both somatic, lipopolysaccharide-dependent phage adsorption and serogroup-specific flagellotropic phages have been demonstrated in other species (16, 26). However, there are no wholly exclusive associations between any phage group and any given O/HL serovar, indicating that phage lysis patterns do vary independently of serogroup. For example, phage group 69 is prevalent among serovar O 23/HL 5 strains but also includes two strains of serovar O 2/HL 4 and one serovar O 50/HL 7 strain.

The biotype markers are more heterogeneous than the phage groups within conserved cell lines. Selective pressure on individual strains as a result of human or veterinary antibiotic usage may account for some of the resistotype variability within serovars. Selection may be particularly important if only single point mutations are necessary to alter any of the five antibiotic resistance characteristics of the Preston biotyping scheme. Plasmid- or phage-mediated transfer of resistance determinants can cause independent variation in resistotypes, and plasmid carriage may also influence phage susceptibility (40). The relatively labile nature of markers from both the Preston biotype and phage type schemes results in their low correlation with clonality (2) and with other strain markers (30), although their high discrimination indices substantiate their usefulness for primary epidemiological investigations.

Clonal associations have been well characterized within several somatic (O-antigen) and flagellar (H-antigen) serovars of the gut pathogen Salmonella enterica (36, 37). Similarly, the polyphasic analysis of strains from three C. jejuni O/HL serovars shows that these groups represent cell lines of common descent. For each serovar, strains sharing genetic and phenotypic characteristics have been isolated from a variety of sources, at different time periods, and from unrelated geographical locations within the UK and abroad. A recent study using flagellin gene typing (25) showed that allelic variation at the flaA locus for common O and HL serogroups of C. jejuni in the United States segregates substantially in accordance with the serovar associations reported here. This provides further evidence for clonal relationships within O/HL serovars. Two serovars which could not be identified in our study due to the use of a restricted set of HL antisera are evident from these data (25). Serogroup O 15 shows clear linkage to HL 13 (flaA type 30), and serogroup O 11 is linked to HL 40 (flaA type 27). The subsidiary association of flaA type 27 with a second O 11 serovar, O 11/HL 8, is also supported by our results (Table 1, row 7).

The clonal groupings identified here embrace strains which do show some differences in either genetic or phenotypic characteristics. Such variable isolates still have a lineage in common, and strains which show broad overall identity but limited variations in phenotype or genotype may more accurately be described as forming a clone complex or clonal subgrouping (3). A similar descriptive was proposed in a recent pulsed-field gel electrophoresis study of HS 1 isolates of C. jejuni (29) to classify strains which showed minor RFLPs with otherwise identical macrorestriction profiles.

As with S. enterica, there is evidence of polyphyletic variation within C. jejuni serovars. Two isolates of serovar O 2/HL 4 (Table 2, strains P/2255 and P/3546) and two of serovar O 50/HL 7 (Table 3, strains P/3758 and P/2578) appear otherwise to be unrelated to the clonal strains of these serovars. Major intragenomic rearrangements of DNA (11, 20) or the chromosomal integration of lysogenic phage (21) could generate this type of atypical restriction pattern diversity within serovars. However, horizontal transfer of serogroup determinants between strains of different genetic backgrounds has been proposed as the primary mechanism for polyphyletic variation in Salmonella (36, 37) and may also be significant in Campylobacter (6, 42). Lateral gene transfer could also account for the replacement of the serovar-characteristic HL 7 antigen with HL 1 and HL 20 epitopes in strains C/13383 and C/19584 (Table 3). Lastly, the possibility of convergent evolution of similar antigens in different cell lines cannot be excluded. Interestingly, the relationship between the O 50 and O 65 serogroup strains reported here (Table 3) is borne out by the results of macrorestriction profile data for these two isolates (13). Evidently, strains with O antigens that do not serologically cross-react can be genotypically closely related.

A recent study of campylobacter population genetics using multilocus enzyme electrophoresis predicted a moderately high frequency of intraspecific recombination for C. jejuni and C. coli, implying that these species should not be clonal (5). However, a clonal population structure has recently been proposed for C. coli (38), and in this report, we provide evidence that serovars represent lineages of common descent in C. jejuni. The apparent absence of genetic barriers to intraspecific recombination in C. jejuni and C. coli (5) suggests that the provenance of such clonal groupings could be recent, possibly due to the rapid proliferation and spread of specific pathogenic strains (23). Large-scale point source outbreaks (24) are one mechanism by which this type of spread could occur. Additionally, modern intensive methods of animal husbandry have been strongly implicated in the recent emergence of several zoonotic pathogens.

In this study, we have demonstrated that many human isolates of C. jejuni show strong associations between specific O and HL antigens and that most strains within three common serovars share genetic and phenotypic characteristics. Further investigations are necessary to characterize the extent of clonal relationships within these and other campylobacter serovars. Where genotypic identities are substantially conserved, rapid and simple phenotypic methods may be the most appropriate means for subtyping within serovars. We conclude that serogrouping with combined O and HL antigens represents both a practical and a phenologically valid method for the epidemiological typing of sporadic isolates of C. jejuni.