Bacillus anthracis virulence in Guinea pigs vaccinated with anthrax vaccine adsorbed is linked to plasmid quantities and clonality

Abstract

Bacillus anthracis is a bacterial pathogen of great importance, both historically and in the present. This study presents data collected from several investigations and indicates that B. anthracis virulence is associated with the clonality and virulence of plasmids pXO1 and pXO2. Guinea pigs vaccinated with Anthrax Vaccine Adsorbed were challenged with 20 B. anthracis isolates representative of worldwide genetic diversity. These same isolates were characterized with respect to plasmid copy number by using a novel method of quantitative PCR developed for rapid and efficient detection of B. anthracis from environmental samples. We found that the copy numbers for both pXO1 and pXO2 differed from those in previously published reports. By combining the data on survival, plasmid copy numbers, and clonality, we developed a model predicting virulence. This model was validated by using a randomly chosen set of 12 additional B. anthracis isolates. Results from this study will be helpful in future efforts to elucidate the basis for variation in the virulence of this important pathogen.

FIG. 1.

Mean ratios of plasmids (pXO1 and pXO2) to chromosomes and percentage of survival of AVA-vaccinated guinea pigs by MLVA clusters. There is a wide range of variability for all of the means across MLVA clusters, and it is difficult to detect any apparent associations between mean plasmid ratios and percentage of survival. The dendrogram below the graph is adapted from previous published genetic typing results (15) and indicates the genetic relationships among the MLVA clusters shown.

FIG. 2.

MLVA cluster factor model of virulence parameter estimates and their 95% confidence intervals (C.I.), along with groupings as indicated by post-hoc testing. MLVA cluster A3b was the arbitrarily set referent factor level and therefore is set at 0. Groups 1 and 2 are overlapping, indicating that there may be as few as three statistically significant different virulence groups or as many as five.

FIG. 3.

Observed and predicted survival rates of AVA-vaccinated guinea pigs challenged with diverse B. anthracis isolates. Twenty isolates were used to construct the model of virulence (A), and a set of 12 isolates was then used to validate the model (B). While there is some deviation between the predicted and observed percentages of survival of guinea pigs when an MLVA cluster factor and plasmid ratio covariates were used, the model is surprisingly robust in performance.

FIG. 4.

Effect of pXO2/chromosome ratio on the percentage of survival across MLVA clusters. Each of the parallel lines shown represents virulence model performance (percentage of survival) for each MLVA cluster while rpXO1 was held constant and rpXO2 varied over the range observed. The figure dramatically illustrates the difference between the predicted percentages of survival for each MLVA cluster, as well as the more subtle effect of additional copies of pXO2 plasmids, which are not otherwise observable.