Materials and Methods

Bacterial isolates.C. jejuni reference strains were obtained from American Type Culture Collection, Culture Collection University of Göteborg (CCUG), and National Collection of Type Cultures. Clinical and animal isolates were identified by use of conventional phenotypic methods, and additional tests for some isolates included hippuricase (hipO) and species-specific polymerase chain reaction (PCR) [5]

MLEEMLEE was done, as described elsewhere [6]. Lysates were prepared from bacterial cultures grown on cefoperazone-vancomycin–amphotericin B blood agar (BD Bioscience) at 42°C under microaerobic conditions and were stored at −70°C until use. We assessed allelic variation for the following enzymes: malic dehydrogenase (MDH), isocitric dehydrogenase (IDH), adenylate kinase (ADK), peptidase of l-phenylalanyl-l-leucine (PEP), aconitase (ACO), malic enzyme, fumarase (FUM), alkaline phosphatase (ALP), and catalase (CAT). Threonine dehydrogenase (THD) was used initially for analysis in our study. However, more than half the strains lacked activity; therefore, we did not include this in our analysis. Buffer system A was used to assay MDH, ADK, and PEP; buffer B was used for IDH, ALP, and ACO; and buffer F was used for ME, FUM, CAT, and THD

Other studiesRFLP analysis of the C. jejuni flaA gene was done, as described elsewhere [4]. Automated ribotyping was performed by use of automated ribotyping, using the RiboPrinter (Qualicon), as described elsewhere [7]. Pulsed-field gel electrophoresis (PFGE) analysis with SmaI and SalI enzymes was done as described elsewhere [7]. The presence of the GM1-like epitope was detected by cholera toxin binding, as described elsewhere [8]

Statistical analysisGenetic diversity, allele frequencies, and genetic distances were calculated by use of ETDIV (version 2.2; http://www.bio.psu.edu/People/Faculty/Whittam/Lab/programs/). Analyses were done initially with null alleles ignored. However, this resulted in several ambiguous assignments to electrophoretic types (ETs). Since the goal of this study was to type organisms that are closely related within 1 species, we considered nulls as a separate allele. Index of association (I_A) was determined by use of MLEECOMP (Ruiting Lan; University of Sydney). Phylogenetic analysis was done by use of parsimony software (PAUP, version 4.0; Sinaeuer Associates)

Results

Eighty-three isolates of C. jejuni were analyzed at 9 different enzyme loci. Multilocus diversity analysis identified 34 ETs in the entire population. For HS:19, there were 17 ETs among 64 isolates, compared with 17 ETs among 19 isolates for the non-HS:19 strains. For control purposes, we also included 4 C. coli isolates; there were 4 ETs in this group (table 1). ETs were rarely shared between the HS:19 and non-HS:19 population. ET4, the most common ET represented in the population (n=45), contained strains with the HS:19 serotype and 1 HS:10 isolate (CCUG 10943). ET6 (n=2) contained 1 HS:19 isolate (HB93-13) and 1 HS:36 isolate (CCUG 10966)

View larger version:

• In this window
• In a new window

• Download as PowerPoint Slide

Figure 1

Phylogenetic analysis of Campylobacter jejuni on the basis of multilocus data. Left 3 Main clusters are outlined with dashed lines. Electropherotype (ET) and heat-stabile (HS) serotypes are shown. Unless indicated by no. in parentheses, ET contains a unique isolate. Details about strains within each ET are described in table 1. ET4 accounted for most HS:19 isolates in cluster I. ETs with an asterisk include isolates from patients with Guillain-Barré syndrome (GBS). Right Dendrogram of automated ribotyping patterns of strains belonging to ET4. Types 1 and 5 accounted for 39 (93%) of 42 strains analyzed. Ribotype band patterns for each strain are shown. Underlined strains are from patients with GBS. Scale indicates percentage of similarity. C. coli, Campylobacter coli; RG, RiboGroup designation

View larger version:

• In this window
• In a new window

• Download as PowerPoint Slide

Table 1

Characteristics of 87 Campylobacter jejuni isolates analyzed by multilocus enzyme electrophoresis

The group of strains comprising HS:19 was less diverse, compared with the group of strains comprising non-HS:19. ET diversity of the HS:19 group was 0.512 versus 0.927 for the non-HS:19 group. ET diversity estimates the probability that 2 isolates chosen at random from within a population are different ETs. The genetic diversity across 9 alleles for 34 ETs ranged from 0.379 for ADK to 0.805 for ALP, with the total diversity being 0.656. For all isolates, genetic diversity was low for HS:19 (0.093±0.022), compared with that for the non-HS:19 population (0.501±074). However, when considering only ETs, diversity was increased to 0.280±0.052 for HS:19, compared with 0.519±0.073 for the non-HS:19 group

The results of phylogenetic analysis are shown in the dendrogram in figure 1. The isolates analyzed in this study could be grouped into 3 major clusters. Cluster I (n=66) contained all of the HS:19 isolates, as well as 2 non-HS:19 strains. A single clone consisting of 45 isolates, which was identified as ET4, was the predominant type in this group. Overall, there were 17 ETs within this cluster. Cluster II consisted of 7 isolates representing 7 ETs and 4 different HS serotypes. Cluster III contained 10 isolates with 8 ETs and represented 9 HS serotypes. Neither cluster II nor III contained any HS:19 isolates

For comparative purposes, 4 strains of C. coli were included in the analysis and diverged from the other strains, a finding which is consistent with the ability of MLEE to differentiate C. jejuni from C. coli [6]. These strains were negative by hipO PCR analysis. Three of these isolates also were serotype HS:19, and 1 was serotype HS:37

A number of additional molecular methods were used to examine the clonality of HS:19 isolates. RFLP analysis of the flaA gene showed that Fla-21 was the most common Fla type among HS:19 strains. Within ET4, 42 (93.3%) of 45 strains were Fla-21. Fla-21 also occurred in the other ETs within cluster I, as did 4 other Fla types (Fla-48, Fla-64, Fla-70, and Fla-110). Automated ribotyping analysis of ET4 isolates also confirmed restricted diversity, with 2 closely related RFLP patterns (types 1 and 5) predominating (figure 1, right panel). These types only differed in 2 band positions. PFGE analysis also showed a similarity of SmaI and SalI profiles within 3 bands of difference (data not shown)

Significant linkage disequilibrium was detected for all enzyme loci, except between PEP and ADK and between CAT and FUM/PEP. The I_A, which was used as a measure of linkage disequilibrium, was 2.738 (±0.145) when the total population was considered but was reduced to 0.695 (±0.231) when only ETs were considered in the analysis. Both values are significantly different from zero and are indicative of a clonal structure

Strains associated with GBS were distributed throughout the MLEE dendrogram and were mixed with strains from patients with only Campylobacter enteritis and with isolates of known animal origin. The same picture was found with respect to the other methods included in this study. The presence of ganglioside-like epitopes in C. jejuni is hypothesized to be involved in the pathogenesis of Campylobacter species–induced GBS. We examined all of the isolates used in this study for the expression of a GM1-like epitope, as determined by their ability to bind cholera toxin. However, the GM1 expression was not associated specifically with GBS strains analyzed in this population

Discussion

Clonal analysis, using MLEE, of a worldwide C. jejuni strain collection showed that HS:19 forms a cluster distinct from other HS serotypes, and a major clone, ET4, was identified within HS:19. Consistent with the clonal nature of ET4 was the observation that most strains were of a single Fla type, FlaA-21, and ET4 was assigned primarily to 2 RiboGroups. The results of this study show that HS:19 strains do not form a monomorphic clone, as suggested by previous investigations [3, 9]. Seventeen closely related ETs containing the HS:19 serotype were observed

The data suggest that ET4 is a worldwide clone, as determined on the basis of identical allelic profiles among strains separated by time and geographic distribution. ET4 contained strains from China, Japan, South Africa, Denmark, United States, Canada, and Mexico. For example, of the 10 isolates from Japan, 5 were from patients with gastroenteritis and were isolated from 1981 through 1983 from 4 different regions in Japan. These strains belonged to the same ET as the 5 strains that were isolated from patients with GBS in 3 regions of Japan from 1987 through 1993. All these isolates had MLEE patterns identical to those of GBS-associated strains isolated in 1997 and 1998 in Mexico City and from 1993 through 1997 in China

Of particular note was the observed relatedness of 2 cluster II HS:41 isolates, DVL5671/ET56 and DVL5724/ET57, with 2 cluster I isolates, INP23/ET7 (HS:19) and INP59/ET23 (HS:41), from patients with GBS in Mexico City. These results suggest that HS:41 and HS:19 strains may have a close common ancestor. HS:41 is the predominant serotype associated with GBS in Cape Town, South Africa [10], and recent studies also suggest a clonal structure for this particular serotype [11]. The observation that HS:19 and HS:41 are closely related also may suggest that these 2 serotypes share some common virulence factor(s) that may be involved in GBS pathogenesis

Aeschbacher and Piffaretti [6] studied a random set of Campylobacter isolates by using MLEE without the knowledge of serotype and demonstrated that Campylobacter species are a largely diverse population with no noted clonal structure. Linkage disequilibrium was not apparent in their collection of isolates, which suggests a random distribution of alleles [6]. Similar results were derived from a study of 156 isolates by Meinersmann [12]. Both studies support the concept that C. jejuni as a population is extremely diverse, yet the index of association found by Meinersmann (I_A=1.29 [±0.212]) suggests clonal development in C. jejuni [12]. Recent multilocus sequence data on a diverse collection of Campylobacter isolates also indicated a weakly clonal population structure [13]

The population structure of Campylobacter species may be similar to that described for Neisseria meningitidis [14]. N. meningitidis exhibits an epidemic population structure in which there appears to be frequent recombination within members of the population, but a successful population occasionally arises and rapidly increases in frequency to produce an epidemic clone [14]. Several studies on the clonal relationships in specific pathogenic serotypes of E. coli have suggested results similar to ours [15]. Both enteropathogenic E. coli and enterohemorrhagic E. coli fall into 2 clonal groups that also have distinct virulence traits. Campylobacter species are known to be naturally competent, and there is evidence for genetic recombination within the species [16]. Except for HS:19, isolates within other serotypes were seen throughout the dendrogram, which suggests horizontal spread of genes involved with the HS serotype (recently identified as capsular polysaccharide genes) not linked to LOS [17]. In contrast, strains within the HS:19 clone complex somehow retained the ancestral serotype genes during expansion of the clone. The origins of this clone and the mechanism of spread throughout the world is unknown

There did not appear to be a link between disease association and ET. Within HS:19, 13 (72.2%) of the 18 GBS-associated strains were ET4, with the remaining 5 isolates each belonging to a different ET. Although this suggests that the ET4 clone is overrepresented by GBS-associated strains, ET4 also contained isolates only associated with diarrheal illness, as well as strains isolated from animals. Although MLEE has been instrumental in identifying pathogenic clones within bacterial populations, such as in E. coli, Haemophilus influenzae, Salmonella species, and N. meningitidis [18], MLEE analysis did not identify a pathogenic clone specifically associated with the development of GBS

The identification of virulence factors that contribute to the pathogenesis of Campylobacter infections are poorly understood mainly because of the lack of good in vitro and in vivo models of infection. There is very good evidence that ganglioside mimicry within the core LOS of Campylobacter is involved in the pathogenesis of GBS. Further studies on the mechanisms of ganglioside mimicry and development of relevant animals models will hopefully shed light on the pathogenesis of Campylobacter-induced GBS. To form a model of this intriguing disease association, a better understanding of the genetic basis of pathogenesis in HS:19-related GBS is needed, as are studies on the role of host factors in immune susceptibility to GBS