Estimation of uncertainty and variability in bacterial growth using Bayesian inference. Application to [i]Listeria monocytogenes[/i]
Résumé
Several models have shown that virulence varies from one strain of Listeria monocytogenes to another, but
little is known about the cause of low virulence. Twenty-six field L. monocytogenes strains were shown to be of
low virulence in a plaque-forming assay and in a subcutaneous inoculation test in mice. Using the results of
cell infection assays and phospholipase activities, the low-virulence strains were assigned to one of four groups
by cluster analysis and then virulence-related genes were sequenced. Group I included 11 strains that did not
enter cells and had no phospholipase activity. These strains exhibited a mutated PrfA; eight strains had a
single amino acid substitution, PrfAK220T, and the other three had a truncated PrfA, PrfA 174-237. These
genetic modifications could explain the low virulence of group I strains, since mutated PrfA proteins were
inactive. Group II and III strains entered cells but did not form plaques. Group II strains had low phosphatidylcholine phospholipase C activity, whereas group III strains had low phosphatidylinositol phospholipase C activity. Several substitutions were observed for five out of six group III strains in the plcA gene and for one
out of three group II strains in the plcB gene. Group IV strains poorly colonized spleens of mice and were
practically indistinguishable from fully virulent strains on the basis of the above-mentioned in vitro criteria.
These results demonstrate a relationship between the phenotypic classification and the genotypic modifications
for at least group I and III strains and suggest a common evolution of these strains within a group.