INTRODUCTION
Campylobacter jejuni and Campylobacter coli are the most commonly identified bacterial agents of gastroenteritis in the developed world [Reference Friedman, Nachamkin and Blase1]. The majority of Campylobacter infections appear to be sporadic with relatively few outbreaks being reported, although this may be due to inadequate ascertainment of related cases [Reference Gilpin2]. The predominant source of the infectious agent is thought to be food. However, the widespread distribution of Campylobacter spp. in the environment suggests that other sources may be important.
New Zealand has one of the highest rates of campylobacteriosis recorded in the developed world [3]. The notification rate in 2006 was 383·5/100 000, which is at least threefold higher than seen in other industrialized countries [Reference Eberhart-Phillip4]. The causes of the increased incidence in New Zealand have yet to be identified, although, as found elsewhere in the world, poultry consumption has been identified as a risk factor [Reference Eberhart-Phillip4–Reference Wingstrand8]. The seasonality and age distribution of cases are similar to those found elsewhere [3, Reference Hudson9, Reference Hearnden10]. To identify potential sources of contamination and transmission routes of Campylobacter spp., the distribution of individual strains within the environment and the relationship between strains must be determined. Attempts to do this in New Zealand using pulsed-field gel electrophoresis (PFGE) have suggested possible transmission routes such as animal faeces and water but definitive answers require a more detailed knowledge of the population structure [Reference Devane11].
The characterization of Campylobacter populations has undergone significant improvements in recent years with the development of a number of genotypic methods [Reference Wassenaar and Newell12]. Many studies have utilized variation in macrorestriction profiles or flagellin gene (flaA) sequence to identify possible outbreaks and examine sources of infection [Reference Harrington, Thomson-Carter and Carter5, Reference Kuusi13, Reference Rivoal14]. However, the large degree of genetic variation seen in both the PFGE and flaA sequence profiles has limited the application of these methods for population studies of C. jejuni and C. coli. The development of a highly portable multi-locus sequence typing (MLST) scheme for C. jejuni and C. coli has provided significant information on the population structure of C. jejuni and C. coli isolates in human clinical samples, animal hosts and the environment [Reference Colles15–Reference Schouls18]. A limited number of clonal complexes (CC) have been detected and host-specific sequence types (ST) and alleles identified. The current C. jejuni/C. coli MLST database containing strain and sequence information is composed mainly of isolates from a limited number of countries (http://pubmlst.org/campylobacter/).
During May and June of 2006 New Zealand experienced an increase in the number of reported campylobacteriosis cases [19]. This unexpected rise in cases observed by the majority of New Zealand district health boards (DHBs) (Fig. 1) prompted this investigation. In total, 112 human isolates of Campylobacter were collected from eight DHBs within a 2-month period and characterized by PFGE, MLST and Penner serotyping.
Fig. 1. The number of campylobacteriosis reports in the district health boards of New Zealand, May 2003–2006. C&C, Capital & Coast; BOP, Bay of Plenty; ▲, 2003; ■, 2004;◆, 2005; □, 2006.
MATERIAL AND METHODS
As part of the investigation into the increase in Campylobacter reports, isolates were requested from laboratories serving the major New Zealand DHBs and eight responded (Fig. 2). All isolates came from diarrhoeic patients (60 females, 52 males) with ages ranging from 0 to 93 years (Table 1). DHBs were asked to submit consecutive isolates excluding repeat isolates and isolates from families. A total of 112 Campylobacter isolates were characterized (five C. coli and 107 C. jejuni). All isolates were grown on 5% (sheep blood) Columbia Blood agar plates at 42°C for 48 h in microaerophilic conditions and speciated following standard microbiological procedures.
Fig. 2. The location of the eight district health boards in New Zealand that sent Campylobacter isolates for typing and subtyping analysis.
Table 1. Subtyping data from New Zealand isolates (isolates are ordered by sequence type)
DHB, District health board; ST, Sequence type; CC, clonal complex; N, North Island; S, South Island; C&C (N), Capital & Coast DHB; BOP (N), Bay of Plenty DHB; UA, unassigned; UT, untypable.
MLST was performed as described previously [Reference Dingle16]. Chromosomal DNA was prepared from freshly grown cultures by boiling for 10 min followed by centrifugation of the disrupted cells. The supernatant was decanted to a fresh tube and used for amplification. The amplifications were performed in a 25 μl volume reaction using Applied Biosystems AmpliTaq Gold mastermix (Applied Biosystems, Auckland, New Zealand) and 5 pmol of each primer. Products were sequenced on an ABI 3130XL automated DNA sequencer using ABI BigDye v3.1 (Applied Biosystems) following the manufacturer's instructions. Sequence data was collated and alleles assigned using the Campylobacter PubMLST database (http://pubmlst.org/campylobacter/). Novel alleles and sequence types were submitted for allele and sequence-type designation as appropriate.
PFGE analysis was performed as described previously [Reference Ribot20]. Isolates were prepared and digested with SmaI and run under standard PulseNet conditions. Salmonella Braenderup H9812 strain restricted with XbaI was run as a size standard. Patterns were clustered using BioNumerics v4.6 (Applied Maths, Ghent, Belgium). A second restriction enzyme, KpnI, was used for further discrimination of isolates with indistinguishable SmaI patterns. PFGE clusters were defined using the BioNumerics software at 95% similarity using an optimization of 0·5% and a position tolerance of 1·5%. Patterns were compared with those in the PulseNet Aotearoa New Zealand Campylobacter database containing 1600 SmaI PFGE patterns and 505 KpnI PFGE patterns from isolates obtained from diverse sources throughout New Zealand since 2001.
The sequence types identified by MLST were assigned to clonal complexes using the eBURST3 programme [Reference Feil21, Reference Spratt22]. Isolates were defined as belonging to a clonal complex if they shared four or more alleles with the central or founder sequence type. Penner serotyping was performed using a panel of 43 C. jejuni antisera produced in-house according to the method of Penner & Hennessy [Reference Penner and Hennessy23].
RESULTS
The 107 C. jejuni and five C. coli isolates received were characterized using PFGE, MLST and Penner serotyping. MLST analysis identified 25 sequence types including four that had not previously been recognized and another three that had been identified in New Zealand isolates only. One new aspA allele was identified in three C. coli isolates (ST-2397) (Table 1). Over one quarter of the isolates were ST-474 (32/112 isolates), 18 were ST-190 and nine were ST-354. ST-190 isolates were identified in all DHBs and ST-474 isolates in seven of the eight DHBs (Table 2). Over half the isolates (59%) belonged to two clonal complexes, CC ST-21 (32 isolates) and CC ST-48 (34 isolates).
Table 2. Sequence types (ST) identified in individual district health boards (DHBs)
BOP, Bay of Plenty DHB; C&C, Capital & Coast DHB.
The PFGE typing data identified isolates from throughout New Zealand with indistinguishable PFGE patterns (Table 1). Within the 112 isolates, 17 PFGE groups of two isolates or more (81 isolates) were identified, and of these 17 groups, 15 had isolates from two or more DHBs (74 isolates). PFGE analysis showed that 23 of the 32 ST-474 isolates belonged to three clusters (cluster J 7 isolates, cluster K 9 isolates and cluster L 7 isolates) with isolates from multiple DHBs. Digestion of the isolates with a second enzyme KpnI indicated that isolates within these clusters were indistinguishable. Similarly, among the 18 ST-190 isolates, three clusters of 10, 3 and 3 isolates were identified (Table 1). Again, digestion with the second enzyme showed that isolates within each cluster were indistinguishable. Comparison with the PulseNet Aotearoa New Zealand Campylobacter database showed that six of the 17 SmaI clusters (B, K, L, M, P and S; Table 1, Fig. 3) had patterns not previously seen in New Zealand. Comparison of the SmaI PFGE patterns from each of the clusters showed that the CC ST-48 patterns were more closely related to each other than to other cluster patterns, as were the CC ST-21 patterns (Fig. 3).
Fig. 3. The PFGE patterns of the 17 clusters identified using SmaI. The KpnI patterns are also shown. ST, Sequence type; CC, clonal complex; UA, unassigned.
Penner serotyping identified 15 different serotypes with the four-complex accounting for 32% of the isolates. All ST-190 isolates except one (untypable) belonged to serogroup 2 and all ST-474 isolates belonged to the four-complex serogroup. Isolates with identical sequence types and indistinguishable PFGE patterns using both restriction enzymes had identical serogroups (Table 1) except for one of the 10 isolates from ST190 cluster F.
DISCUSSION
Outbreaks of campylobacteriosis are rarely identified in relation to the number of cases reported. In 2005, 13 839 cases of campylobacteriosis were reported in New Zealand, yet only 47 Campylobacter-associated outbreaks were reported involving 252 cases [3]. This suggests either a significant under-reporting of outbreaks or a very large number of sporadic cases. Unfortunately the volume of Campylobacter isolates normally precludes routine subtyping for identification of related cases and may hide the true extent of outbreak-related cases. Recent studies using MLST and PFGE subtyping methods have shown that within human Campylobacter strains there is a clear temporal distribution of isolates [Reference Gilpin2, Reference Mickan24, Reference Sopwith25]. In New Zealand, PFGE subtyping of all isolates received by a single clinical laboratory over two short time periods identified a significant number of clusters [Reference Gilpin2]. Distinct clusters defined using both SmaI and KpnI PFGE patterns were identified in different isolation periods suggesting outbreaks may be more common than previously thought. Similarly, studies characterizing isolates by MLST, one from urban and rural communities in the United Kingdom [Reference Sopwith25] and one from New South Wales, Australia [Reference Mickan24], also identified a temporal distribution of clonal complexes. The lower discriminatory power of MLST precludes identification of potential outbreaks per se but clearly indicates clusters of related isolates present within a collection. In this study we have identified distinct clusters within a discrete time-frame but from isolates distributed across the length of New Zealand.
Two restriction enzymes (SmaI and KpnI) were used for the confirmation of individual clusters and MLST was included as an alternative confirmatory technique providing unequivocal data on the relationship of the isolates. It has previously been shown that isolates from a number of confirmed outbreaks were indistinguishable by four subtyping methods including MLST, PFGE and Penner serotyping [Reference Sails, Swaminathan and Fields26]. Isolates in all the major clusters (Table 1) were indistinguishable by PFGE and MLST and Penner serotyping, although one isolate in ST-190 cluster F had a different serotype. As found previously ST-257 was only identified with serotype 11 and ST-42 with serotype 23,36 [Reference Dingle27].
The sudden and unexpected increase in Campylobacter isolates during the winter of 2006 [19] provided an opportunity to examine the relationship and distribution of Campylobacter subtypes across New Zealand. This increase was specific to New Zealand and not identified in Australia. The 112 isolates were obtained from eight DHBs, four on the North Island and four on the South Island (Fig. 2). The surprising result was the identification of indistinguishable isolates from upwards of five different DHBs located on both islands (Table 1). This could be attributed to a generalized increase in exposure to multiple sources associated with a common risk factor, or a common source outbreak arising from a single source that was widely distributed throughout New Zealand. The relative frequency and spatial pattern of the genotypes, especially the MLST types, are more consistent with the latter scenario: such a marked increase in notification, accompanied by the predominance of indistinguishable strains with a wide spatial distribution, is consistent with a common source epidemic. Strains most likely associated with a common source are ST-474 and ST-190; these were the most prevalent sequence types, with ST-190 identified in all regions and ST-474 in seven of the eight regions, and the only sequence types isolated in the Lakes region (Table 2). Unfortunately the small numbers associated with each cluster meant that no significant information on possible sources of infection could be identified from epidemiological information gathered and the rural or urban nature of the sample was not recorded.
Comparison of the PFGE patterns from the two major sequence-type groups with those in the PulseNet Aotearoa Campylobacter database showed that the PFGE patterns of two clusters had not been seen previously (ST-474 cluster K and cluster L), one had been seen once (ST-190 cluster D), one four times (ST-474 cluster J), one 23 times (ST-190, cluster F) and one 44 times (ST-190 cluster E). It is clear from the PFGE and MLST data (P. Carter and S. McTavish, unpublished data) that particular Campylobacter strains are very stable over a number of years and continue to cause human infections. This stability obfuscates the relationship of these strains in outbreak scenarios and further work surrounding their epidemiology is required. Identification of stable strains within Campylobacter populations using MLST, PFGE, flaA RFLP typing and AFLP has been reported previously among human and poultry isolates [Reference Harrington, Thomson-Carter and Carter5, Reference Manning28, Reference Manning29].
One unique strain, ST-474 cluster K, was identified in four different DHBs and the PFGE patterns had not previously been seen. The widespread occurrence of this strain argues against a simple local point source normally associated with Campylobacter outbreaks but may reflect the need to look at other possible reasons such as food distribution within New Zealand. This scenario is comparable to the widespread dissemination of E. coli O157:H7 through large-scale food distribution networks in the United States and highlights the importance of subtyping and surveillance (e.g. by PulseNet USA) in identifying such outbreaks.
The MLST data identified a number of commonly described sequence types that have previously been associated with human infection. CC ST-21 has been identified in isolates from a wide range of sources accounting for up to a third of human isolates (20–33%) [Reference Schouls18, Reference Sopwith25, Reference Dingle27, Reference Karenlampi30] which was also the case in this study. The other major clonal complex in this study, CC ST-48, also accounted for approximately one third of human isolates, comparable with recent data from Australia [Reference Mickan24], although significantly more than reported previously in other studies (5–10%) [Reference Schouls18, Reference Sopwith25, Reference Dingle27, Reference Karenlampi30]. The majority of the CC ST-48 isolates in New Zealand were ST-474, which is not a commonly identified sequence type internationally. Only one isolate with this sequence type, a Czech isolate from chicken, is logged in the MLST database. Interestingly, the Czech Republic is also reported to have a very high rate of Campylobacter infection [31]. It has not been reported in other MLST studies of Campylobacter isolates. It is present, however, in significant numbers (about 10% of samples characterized) in human and poultry isolates in New Zealand and has been isolated from sheep and cows (N. French and P. Carter, unpublished results). It may be that this particular sequence type is endemic in New Zealand but not prevalent elsewhere in the world. Potentially endemic strains have also been identified in Australia [Reference Mickan24] and Curacao [Reference Duim32]. There were also four isolates with sequence types that had been previously identified in New Zealand as novel sequence types, associated with chicken meat (ST-2343, one isolate and ST-2345, two isolates) and river water (ST-2347, one isolate). The SmaI PFGE pattern of the ST-2343 and ST-2347 isolates were indistinguishable from those seen previously. The two ST-2345 isolates in this study, however, gave different patterns to those seen previously. These sequence types may represent other endemic strains of Campylobacter.
The data presented here regarding the unusual increase in campylobacteriosis in New Zealand over the winter of 2006 are consistent with a common source epidemic associated with endemic strains of Campylobacter. Clones of Campylobacter identified by PFGE patterns, MLST and Penner serotyping are widely distributed throughout New Zealand, some of which have been identified before and represent stable clones. The PFGE patterns associated with individual clonal complexes are closely related, consistent with the distribution of epidemic strains via an unknown source. The identification of the New Zealand endemic strain, ST-474, and its association with food sources and serious human illness warrants further investigation.
ACKNOWLEDGEMENTS
We thank all of the laboratory staff involved from the district health boards for providing the campylobacter isolates and members of the Enteric Reference Laboratory, ESR, for their work. We also thank Dr Ronan O'Toole, School of Biological Sciences Victoria University of Wellington for the co-supervision of Sharla McTavish in her Masters Dissertation. S.M. is a recipient of an ESR Masters scholarship. This work was part funded by the Ministry of Health.
DECLARATION OF INTEREST
None.
