OUP user menu

In vitro genotypic variation of Campylobacter coli documented by pulsed-field gel electrophoretic DNA profiling: implications for epidemiological studies

Stephen L.W. On
DOI: http://dx.doi.org/10.1111/j.1574-6968.1998.tb13167.x 341-346 First published online: 1 August 1998


Six isolates of Campylobacter coli from different pig herds were subcultured up to 50 times over a 6-month period and DNA samples suitable for pulsed-field gel electrophoretic (PFGE) profiling prepared at regular (1, 20, 40 and 50 passages) intervals. In 5/6 strains, changes in the banding patterns of Sma1, Sal1 and/or BamH1 digests were observed. In one such strain the differences were considered artifactual. However, significant alterations in PFGE profiles between subcultures of four strains were seen, irrespective of the restriction enzyme used. Spontaneous intramolecular genomic rearrangements were considered the most likely mechanism for the changes observed. A numerical analysis based upon the combined distribution of Sma1- and Sal1-derived fragments clustered most strain subcultures together, with the exception of those from one isolate which were divided into two clusters. The effect of spontaneous genetic change on PFGE profiles must be considered when evaluating strain relationships. Numerical techniques may aid data interpretation but results must be evaluated cautiously.

Key words
  • Campylobacter coli
  • PFGE
  • Genotypic variation
  • Numerical analysis
  • Evolutionary genetics

1 Introduction

The enteropathogenic Campylobacter species C. jejuni, C. coli, C. lari and C. upsaliensis are considered to be the most commonly isolated bacteria from human diarrhea [1]. Their clinical and economic significance is well documented [1]. The bacteria are widely distributed in nature and transmission is principally believed to be foodborne [1]. However, most reported cases of human infection appear sporadically, making the accurate identification of the sources of disease difficult. It has been suggested that the use of high-resolution genotyping methods could significantly improve our understanding of campylobacter epidemiology [2].

The method of macrorestriction profiling by pulsed-field gel electrophoresis (MRP-PFGE) is considered to be among the most discriminatory molecular epidemiological tools available for bacterial fingerprinting [3, 4]. However, the interpretation of data from MRP-PFGE (and other genotyping methods) for investigating the epidemiology of infections that are detected sporadically, or over extended periods of time, requires special consideration [3]. Moreover, an understanding of the evolutionary genetics of the organism under study is desirable to evaluate the significance of data from molecular epidemiological studies [5].

It has been suggested from a multifactorial genotypic investigation of C. coli that campylobacters do not exhibit a clonal population genetic structure [6]. These data are somewhat controversial [7]. Although campylobacters, particularly C. coli, are able to transform with free DNA [8], several genotyping studies suggest that stable clones can be found from diverse sources [2, 9, 10]. Clearly more data concerning the nature of genetic change in campylobacters would be useful to clarify this issue, and thus verify the validity (or otherwise) of genotyping campylobacters for epidemiological studies of sporadic infection, or for long-term continuous surveillance, as performed in Denmark for zoonotic bacteria.

The aim of this study was to evaluate the extent and nature of any spontaneous genotypic change, as detected by MRP-PFGE, in C. coli that may occur over an extended period of continuous subculturing in vitro. In addition, the validity of numerical analysis of the resulting profiles for determining interstrain relationships was also examined.

(Certain of these data were presented at IMBEM IV, 10–13 September 1997, Elsinore, Denmark, as abstract P16 on p. 79 of the proceedings.)

2 Materials and methods

2.1 Bacterial strains

Six C. coli isolates from different pig herds in Denmark were studied. All strains gave reactions in several key phenotypic tests (namely cell morphology, oxidase, catalase, hippurate hydrolysis, growth at 42°C and on a minimal medium, and α-haemolysis on blood agar) typical of the species [11]. The strains were selected from an initial Sma1 MRP-PFGE screening of 52 porcine C. coli strains and were known to be from different slaughterhouses, as well as genotypically distinct (S.L.W. On and J. Neimann, unpublished data).

2.2 Subcultivation

Strains were initially recovered from deep-frozen stocks by culturing on petri dishes containing 5% bovine blood agar under microaerobic conditions at 37°C and continually subcultured (using one plate per strain), 1–2 times per week, under the same growth conditions for approximately 6 months. During this period, these strains were maintained separately from all other isolates and no other C. coli strains were knowingly studied in the laboratory.

2.3 DNA sample preparation, digestion and electrophoresis

DNA samples suitable for MRP-PFGE were prepared from strains subcultured after 1, 20, 40 and 50 passages (following initial recovery from the frozen state) using methods described previously [10]. Macrorestriction profiles (MRPs) based on digestion with enzymes Sma1, Sal1 or BamH1 and subsequent electrophoresis and visualization were determined using methods and electrophoresis conditions described previously [10].

2.4 Numerical analysis of macrorestriction profiles

Photographs of ethidium bromide-stained gels of Sma1 and Sal1 digests were digitized using a desktop scanner (Scanjet IICX, Hewlett-Packard, California, USA). Negative images of the DNA patterns were optimized and assimilated for numerical analysis by GelCompar version 4.0 (PC-compatible) software. Inter-gel variation was corrected by use of a standard reference pattern (Sma1 digests, CCUG 14169; Sal1 digests, λ ladder). The position of each band in the digitized profiles was defined with careful reference to the original photograph and to the densitometric curve of the scanned image. Optimal band alignment was performed automatically, using a maximum position tolerance of 0.5%. Similarities between strains were based on band polymorphisms of both Sma1 and Sal1 digests and calculated using the Dice coefficient. A dendrogram was constructed to reflect the similarities in the matrix. Strains were clustered using the method of Ward [12].

3 Results

3.1 General features of C. coli macrorestriction profiles

Sma1 digests of the strains examined contained 10–17 bands ranging from ca. 20–630 kb in size (Fig. 1). Sal1 digests were somewhat less complex, containing 6–13 bands ranging from ca. 40–600 kb (Fig. 2). An accurate estimate of the number of fragments comprising BamH1 digests could not be made, due to the complex nature of the patterns (Fig. 3). The same profiles were obtained when DNA samples were reexamined by the same methods (data not shown).


Sma1-derived PFGE-DNA profiles of subcultured C. coli strains and reference strain C. hyointestinalis. Gel A: lanes 1–4, Lab 27 after 1, 20, 40 and 50 passages respectively; 5–6 and 9–10, Lab 33 (after 1, 20, 40, 50 passages); 11–14, Lab 113 (after 1, 20, 40, 50 passages). Gel B: lanes 1–4, Lab 215 (after 1, 20, 40, 50 passages); 5–6 and 9–10, Lab 332 (after 1, 20, 40, 50 passages); 11–14, Lab 421 (after 1, 20, 40, 50 passages). Lanes 7 and 8 on both gels are λ ladder and CCUG 14169T (C. hyointestinalis subsp. hyointestinalis) respectively.


Sal1-derived PFGE-DNA profiles of subcultured C. coli strains and reference strain C. hyointestinalis. Lane identities are identical to those given for Fig. 1.


BamH1-derived PFGE-DNA profiles of subcultured C. coli strains and reference strain C. hyointestinalis. Lane identities are identical to those given for Fig. 1.

3.2 Variation in macrorestriction profiles of strains after 1, 20, 40 and 50 passages

Differences in strain MRPs between subcultures were observed in five of the six strains examined. No discernible differences were detected in MRPs of Lab 33, nor in Sal1 digests of Lab 332 subcultures. The most significant change was noted in Lab 27, where profiles obtained after one and 50 passages shared only six of 18 total band loci for Sma1 digests, and five of 11 band loci for Sal1 digests (Fig. 1A and Fig. 2A respectively); notable genotypic change was also evident in BamH1 profiles (Fig. 3A). Between-subculture differences in the remaining strains were also noted in each of their Sma1, Sal1 and BamH1 profiles (Figs. 13), but the observed variation was less marked compared with Lab 27. Nonetheless, the number of common bands varied between 8–11 of 16–18 band loci for Sma1 digests (Fig. 1); and 6–7 of 12–15 band loci for Sal1 digests (Fig. 2). In general, strain variation between 1–20 and 40–50 passages was limited, with the most notable changes observed between 20 and 40 passages.

3.3 Numerical analysis of Sma1 and Sal1 macrorestriction profiles

Inter-gel reproducibility was determined as 96.7%, which was the percentage similarity between duplicate (Sma1 and Sal1) profiles of reference strain CCUG 14169.

Eight clusters were defined at the 92.5% similarity (S-) level. One of these contained the aforementioned reference strain duplicates, whilst subcultures of five of the six C. coli strains also clustered together. Subcultures of Lab 27 were distributed among two clusters, containing profiles from passages 1 and 20; and 40 and 50 respectively.

4 Discussion

The results obtained here demonstrate that the MRPs of some C. coli strains can change significantly after extended subculturing under standard in vitro culture conditions. In one case (Lab 332) the differences observed between Sma1- and BamH1-derived MRPs of the strain subcultures may be explained by methodological anomalies. It is clear from the intensity of the stained patterns that the DNA content of samples produced at 1 and 20 passages is greater than those made at later stages (Fig. 1, Fig. 2, Fig. 3B). The relatively minor differences in the Sma1 and BamH1 digests of the earlier two samples is likely to be a consequence of this factor alone, given that there are no discernible differences between comparable Sal1 MRPs. However, four of the six strains examined showed clear differences in Sma1, Sal1 and BamH1 MRPs between subcultures. Certain of these differences indicate an evolution in genotype. Since DNA samples were prepared from plate cultures, they are representative of a population of cells that are not necessarily at the same stage of development. A comparison of the Sma1 MRPs of Lab 27 illustrates this point, since subculture 20 displays several features (notably bands at ca. 600, 550, 400, and 40 kb) that suggest the presence of cells containing earlier and later genotypes (Fig. 1A).

The differences between MRPs cannot readily be explained by mechanisms such as point mutation (since more than one restriction site is affected and the banding distribution is not entirely consistent with even multiple events of this type) or transformation (no other donor of DNA present), which are known to alter macrorestriction profiles in bacteria [3, 4]. The changes in the macrorestriction profiles noted here could be explained by multiple recombinational events, either spontaneous genomic rearrangements or recombinations of mobile elements. Intragenomic recombination has been postulated as a means by which genetic profiles based on the flagellin gene can change [13], but such change as is described here would probably involve recombination of several gene loci. Genomic rearrangements could account for the discrepant genotyping results obtained by others [6, 14, 15]. Whilst it could be argued that the prolonged period of subculture used here represents a wholly artificial method for inducing genotypic change, it is noteworthy that spontaneous changes in DNA restriction patterns of C. coli strains have also been described in experimental infections of chicks [14]. In addition, differences noted between macrorestriction profiles of C. jejuni isolates from a single batch of poultry meat were attributed to genomic rearrangements [15].

It is likely that the aforementioned genetic phenomena are not the only mechanisms that may elicit changes in DNA-based fingerprints. Both C. coli and C. jejuni are naturally competent [8] and natural transformation was suggested as an explanation for changes in the macrorestriction profile of an outbreak-associated C. coli strain [16]. Differences between MRPs of an unusual Campylobacter-like organism [17] subsequently shown to be a novel biovar of C. sputorum[18] could be attributed to point mutations in restriction sites. Each of the genetic mechanisms discussed above makes the interpretation of macrorestriction profiles for epidemiological studies difficult using present guidelines [3, 4], which are primarily intended for investigating putative outbreaks. Under such conditions, identification of the ‘outbreak type’ is required for interstrain comparison and spontaneous genotypic change among strains of the outbreak type is assumed to occur infrequently, with the effect on MRPs limited to 1–3 ‘genetic events’ (e.g. single band shift or restriction site mutation) [3, 4]. It is accepted that the aforementioned guidelines are difficult to apply when applied to bacteria yielding MRPs composed of less than 10 fragments, in continuous surveillance studies, or epidemiological investigations of sporadic infections such as campylobacteriosis [3, 4]. However, the present study also shows that genomic rearrangements can result in changes in strain MRPs that cannot be explained by simple mutation or insertion events, and may result in dramatic shifts in the proportion of shared bands between subcultures of the same strain. Additional methods for interpreting molecular typing data for campylobacters would clearly be beneficial. Numerical analysis of MRPs has been used to determine relationships between strains of Aeromonas hydrophila[19], Salmonella typhimurium[20] and C. jejuni[21]. The present study confirms the general usefulness of this method for correctly identifying strain relationships, since it successfully clustered five of the six groups of C. coli subcultures together (Fig. 4). However, the strain in which the most extensive genotypic change was seen (Lab 27) fell into two clusters, indicating that numerical analysis is a useful but certainly not infallible tool for identifying bacterial clones. In any case, the results of such analyses should always be cautiously interpreted [20].


Dendrogram of the cluster analysis of Sma1- and Sal1-derived DNA-PFGE profiles of C. coli and C. hyointestinalis strains. Figures along the horizontal axis indicate % similarities as determined by the Dice coefficient and Ward's clustering. The number of passages (n= 1, 20, 40 or 50) of the C. coli strains are appended to the strain number, e.g. Lab 27-1 is strain Lab 27 after one passage.

The increasing reports concerning genomic instability in campylobacters ([6, 1317], this study) and the consequent effect on molecular fingerprints do not invalidate the use of such methods for investigating complex epidemiological situations such as sporadic campylobacteriosis. Several studies have clearly indicated that genetically stable clones of C. jejuni[2, 10] and C. coli[9] can be identified among sporadic isolates from diverse sources. The present study also demonstrates that not all strains of C. coli show the same propensity for genetic change. All of the aforementioned factors clearly need to be considered when interpreting the results of genotyping studies of these organisms, in order to ensure that a meaningful insight into their epidemiology is obtained. Further studies concerning the reasons why some strains appear to be more genetically stable than others may also be of interest. Nonetheless, present data indicate that genotypically non-identical strains may, in fact, represent the same strain. The implications for epidemiological, and population genetic studies of C. coli at least, are self-evident.


I thank P. Jordan, S. Hewitt and B.H. Christensen for technical assistance.


View Abstract