Comparative genomic assessment of Multi-Locus Sequence Typing: rapid accumulation of genomic heterogeneity among clonal isolates of Campylobacter jejuni
1 Laboratory for Foodborne Zoonoses (Lethbridge Unit), Public Health Agency of Canada c/o Animal Diseases Research Institute, PO Box 640, Township Road 9-1, Lethbridge, Alberta, T1J 3Z4, Canada
2 Laboratory for Foodborne Zoonoses, Public Health Agency of Canada, 110 Stone Road West, Guelph, Ontario, N1G 3W4, Canada
3 Institute for Biological Sciences, National Research Council of Canada, 100 Sussex Drive, Ottawa, Ontario, K1A 0R6, Canada
BMC Evolutionary Biology 2008, 8:229 doi:10.1186/1471-2148-8-229Published: 8 August 2008
Multi-Locus Sequence Typing (MLST) has emerged as a leading molecular typing method owing to its high ability to discriminate among bacterial isolates, the relative ease with which data acquisition and analysis can be standardized, and the high portability of the resulting sequence data. While MLST has been successfully applied to the study of the population structure for a number of different bacterial species, it has also provided compelling evidence for high rates of recombination in some species. We have analyzed a set of Campylobacter jejuni strains using MLST and Comparative Genomic Hybridization (CGH) on a full-genome microarray in order to determine whether recombination and high levels of genomic mosaicism adversely affect the inference of strain relationships based on the analysis of a restricted number of genetic loci.
Our results indicate that, in general, there is significant concordance between strain relationships established by MLST and those based on shared gene content as established by CGH. While MLST has significant predictive power with respect to overall genome similarity of isolates, we also found evidence for significant differences in genomic content among strains that would otherwise appear to be highly related based on their MLST profiles.
The extensive genomic mosaicism between closely related strains has important implications in the context of establishing strain to strain relationships because it suggests that the exact gene content of strains, and by extension their phenotype, is less likely to be "predicted" based on a small number of typing loci. This in turn suggests that a greater emphasis should be placed on analyzing genes of clinical interest as we forge ahead with the next generation of molecular typing methods.