Evolution of an agriculture-associated disease causing Campylobacter coli clade: evidence from national surveillance data in Scotland

Abstract

The common zoonotic pathogen Campylobacter coli is an important cause of bacterial gastroenteritis worldwide but its evolution is incompletely understood. Using multilocus sequence type (MLST) data of 7 housekeeping genes from a national survey of Campylobacter in Scotland (2005/6), and a combined population genetic-phylogenetics approach, we investigated the evolutionary history of C. coli. Genealogical reconstruction of isolates from clinical infection, farm animals and the environment, revealed a three-clade genetic structure. The majority of farm animal, and all disease causing genotypes belonged to a single clade (clade 1) which had comparatively low synonymous sequence diversity, little deep branching genetic structure, and a higher number of shared alleles providing evidence of recent clonal decent. Calibration of the rate of molecular evolution, based on within-species genetic variation, estimated a more rapid rate of evolution than in traditional estimates. This placed the divergence of the clades at less than 2500 years ago, consistent with the introduction of an agricultural niche having had an effect upon the evolution of the C. coli clades. Attribution of clinical isolate genotypes to source, using an asymmetric island model, confirmed that strains from chicken and ruminants, and not pigs or turkeys, are the principal source of human C. coli infection. Taken together these analyses are consistent with an evolutionary scenario describing the emergence of agriculture-associated C. coli lineage that is an important human pathogen.

Figure 1. ClonalFrame genealogy of 7-locus MLST data from C. coli genotypes from clinical, animal faeces, food and environmental samples.

STs from farm only (green), riparian only (blue), clinical only (red) and farm and clinical (black) sources are indicated by different colours.

Figure 2. Maximum likelihood (ML) analysis of congruence in Campylobacter jejuni and Campylobacter coli.

The ML tree of each MLST locus is compared with the ML trees from the other six loci for: (A) C. coli clades 1–3, (B) C. jejuni, (C) C. coli and C. jejuni (D) C. coli clade1, (E) C. coli clade 2 and (F) C. coli clade 3. The tree congruence likelihood (ln L) range is shown with horizontal bars for each locus (black) and for 200 trees of random topology (grey) for aspA, glnA, gltA, glyA, pgm, tkt and unc respectively from top to bottom on each graph. If the ln L range falls outside the range calculated for random trees there is more congruence among the ML trees than expected by chance.

Figure 3. Phylogeny of estimated divergence times in the genus Campylobacter using BEAST.

Error bars, in grey, associated with each node indicate relative uncertainty in node height. Calibration of the scale bar was based on intraspecific variation in Campylobacter jejuni and uncertainty associated with the molecular clock calibration is represented by a 95% CI below the scale bar. Empirical and indirect estimates of the rate of molecular evolution would calibrate the same scale bar at 42,200 (95% CI 2,690–661,000) and 7,600,000 years respectively.