Low dose aerosol fitness at the innate phase of murine infection better predicts virulence amongst clinical strains of Mycobacterium tuberculosis

Abstract

Background: Evaluation of a quick and easy model to determine the intrinsic ability of clinical strains to generate active TB has been set by assuming that this is linked to the fitness of Mycobacterium tuberculosis strain at the innate phase of the infection. Thus, the higher the bacillary load, the greater the possibility of inducting liquefaction, and thus active TB, once the adaptive response is set.

Methodology/principal findings: The virulence of seven clinical Mycobacterium tuberculosis strains isolated in Spain was tested by determining the bacillary concentration in the spleen and lung of mice at weeks 0, 1 and 2 after intravenous (IV) inoculation of 10⁴ CFU, and by determining the growth in vitro until the stationary phase had been reached. Cord distribution automated analysis showed two clear patterns related to the high and low fitness in the lung after IV infection. This pattern was not seen in the in vitro fitness tests, which clearly favored the reference strain (H37Rv). Subsequent determination using a more physiological low-dose aerosol (AER) inoculation with 10² CFU showed a third pattern in which the three best values coincided with the highest dissemination capacity according to epidemiological data.

Conclusions/significance: The fitness obtained after low dose aerosol administration in the presence of the innate immune response is the most predictive factor for determining the virulence of clinical strains. This gives support to a mechanism of the induction of active TB derived from the dynamic hypothesis of latent tuberculosis infection.

Figure 1. Fitness results.

Fitness results from in vivo and in vitro studies. Top: fitness values for lung aerosol and IV experiments; bottom: in vitro assay and spleen IV infection values. Dotted lines indicate the fitness value for the H37Rv strain (shown as “St”).

Figure 2. Fitness in the in vivo models.

Bacillary load in the lung from aerosol-infected mice and lung and spleen from IV-infected mice on day 0, week 1 and week 2. The H37Rv strain is indicated as St. Dotted lines show the inocula used to infect each mouse.

Figure 3. Aggregate area distribution results.

Cording area analysis results for the strains inoculated, as obtained by image analysis. A and B: representation of aggregate area distribution showing the two characteristic patterns found. A: log-normal distribution observed for strain 3 (included in the fast strains group). B: decreasing exponential distribution observed for strain 5 (included in the slow strains group). C: picture from a Ziehl-Neelsen stained strain 1 sample showing different sized cords. Original magnification: 200×.

Figure 4. Division of clinical strains into fast and slow strains.

The strains were divided in fast (1, 2 and 3) and slow strains (4, 5, 6 and 7) according to their fitness in lung in the in vivo model IV (panel A). Panel B and C show the representation of fast and slow strains according to parameters determined from the cording area analysis: skewness (panel B), and proportion of single bacilli (brown circles) and of aggregates (bright red circles) (panel C). A significant difference was found between fast and slow strains for all three graphs.

Figure 5. Fitness and Lag phase results.

Comparison of fitness values from the different experiments with respect to the number of clinical cases detected (A) and the evaluated lag phase(B–D). A: correlation between fitness values from lung aerosol infections and the number of clinical cases reported, which resulted significant; B–D: fitness (left axes, filled circles) and lag phase (right axes, unfilled circles) of each experiment with respect to the different strains in the lung after aerosol infection (B), lung after IV infection (C) and in vitro assay (D).