Paucibacillary Tuberculosis in Mice after Prior Aerosol Immunization with Mycobacterium bovis BCG

Abstract

To develop a murine model of paucibacillary tuberculosis for experimental chemotherapy of latent tuberculosis infection, mice were immunized with viable Mycobacterium bovis BCG by the aerosol or intravenous route and then challenged six weeks later with virulent Mycobacterium tuberculosis. The day after immunization, the counts were 3.71 ± 0.10 log₁₀ CFU in the lungs of aerosol-immunized mice and 3.65 ± 0.11 and 4.93 ± 0.07 log₁₀ CFU in the lungs and spleens of intravenously immunized mice, respectively. Six weeks later, the lungs of all BCG-immunized mice had many gross lung lesions and splenomegaly; the counts were 5.97 ± 0.14 and 3.54 ± 0.07 log₁₀ CFU in the lungs and spleens of aerosol-immunized mice, respectively, and 4.36 ± 0.28 and 5.12 ± 0.23 log₁₀ CFU in the lungs and spleens of intravenously immunized mice, respectively. Mice were then aerosol challenged with M. tuberculosis by implanting 2.37 ± 0.13 log₁₀ CFU in the lungs. Six weeks after challenge, M. tuberculosis had multiplied so that the counts were 6.41 ± 0.27 and 4.44 ± 0.14 log₁₀ CFU in the lungs and spleens of control mice, respectively. Multiplication of M. tuberculosis was greatly limited in BCG-immunized mice. Six weeks after challenge, the counts were 4.76 ± 0.24 and 3.73 ± 0.34 log₁₀ CFU in the lungs of intravenously immunized and aerosol-immunized mice, respectively. In contrast to intravenously immunized mice, there was no detectable dissemination to the spleen in aerosol-immunized mice. Therefore, immunization of mice with BCG by the aerosol route prior to challenge with a low dose of M. tuberculosis resulted in improved containment of infection and a stable paucibacillary infection. This model may prove to be useful for evaluation of new treatments for latent tuberculosis infection in humans.

There is a need for new treatment regimens for latent tuberculosis infection (LTBI) that are easier to administer and active against multidrug-resistant tuberculosis. Because latency is still poorly understood, a limited number of models of LTBI exist for evaluation of new regimens. Mice are the most commonly used animal models for preclinical testing of antimicrobial efficacy, but mice do not develop latent Mycobacterium tuberculosis infections analogous to LTBI in humans. Immunocompetent mice infected by the intravenous route with 5 × 10³ CFU of virulent M. tuberculosis develop a chronic, nonlethal infection in which the counts plateau at approximately 10⁶ CFU per lung or spleen (4). Similar observations have been made for lung counts after low-dose aerosol infection (6). Although the plateaus in counts in both circumstances represent containment of the infections by host immunity, the number of bacilli causing an ongoing infection is more consistent with the size of the bacillary population in human lungs during chronic tuberculosis disease than with the size of the bacillary population during the paucibacillary state associated with LTBI. In the latter case, the size of the bacillary population is assumed to be less than 10⁴ CFU since 10⁴ CFU/ml is the lower limit of detection with acid-fast smears and latent lesions found at autopsy are culture positive but smear negative (2, 14).

There are no murine models of paucibacillary tuberculosis infection resulting from containment by host immune mechanisms. Previous work by McCune et al. (12) gave rise to the so-called Cornell model that has been used to evaluate new drug regimens for LTBI. While the Cornell model is based on creation of a paucibacillary state of culture-negative persistence, this state is created by exposure to antituberculosis drugs and not containment of the initial infection by host immunity. Organisms in a culture-negative state following prolonged antibiotic therapy may be fundamentally different from organisms causing LTBI in humans. Lecoeur and colleagues have introduced a murine model of LTBI in which host immunity is augmented by intravenous infection with Mycobacterium bovis BCG 4 weeks prior to intravenous challenge with virulent M. tuberculosis (9). Improved containment of infection results in M. tuberculosis counts that plateau at 10⁵ rather than 10⁶ bacilli in the lungs and spleen. This model proved its utility when it was used to demonstrate the remarkable sterilizing activity of rifampin and pyrazinamide compared to that of isoniazid (9), which resulted in development of this combination as a highly effective treatment for LTBI in humans. Still, the model is not a model of paucibacillary disease. Its predictive power might be improved by augmentation of the immune response to further restrain bacterial multiplication to achieve a bacillary population of less than 10⁴ organisms.

It has been demonstrated previously that immunization of mice with BCG by the aerosol route elicits better protection against aerosol challenge with M. tuberculosis than immunization by the intravenous and subcutaneous routes elicits (10). However, because a challenge dose of >10⁴ CFU was used in this previous study, the size of the population of tubercle bacilli in the lungs of aerosol-immunized mice approached 10⁵ CFU. We hypothesized that the size of the bacillary population in the lungs could be restricted further by using a low-dose aerosol challenge with M. tuberculosis. Therefore, we designed an experiment to determine if the model of LTBI introduced by Lecoeur and colleagues (9) could be improved to create a paucibacillary infection to better mimic LTBI in humans by using an aerosol immunization strategy and subsequent aerosol challenge with a low dose (500 CFU) of M. tuberculosis. A second aim was to determine if aerosol immunization with BCG is more effective than immunization by the intravenous route in containing the multiplication of M. tuberculosis following a low-dose aerosol challenge rather than a high-dose aerosol challenge.

MATERIALS AND METHODS

Mice.

A total of 132 female BALB/c mice (Charles River, Wilmington, Mass.) that were 6 weeks old were randomly distributed into four groups: (i) a nonimmunized normal control group of 12 mice that was not challenged later, (ii) a nonimmunized control group of 18 mice that was challenged with M. tuberculosis later, (iii) a group of 38 mice immunized with BCG by the intravenous route, and (iv) a group of 38 mice immunized with BCG by the aerosol route. All animal care and experimental manipulations were approved by the institutional animal care and use committee.

Mycobacterial strains.

BCG Pasteur was kindly provided by F. Collins, Food and Drug Administration, Bethesda, Md. M. tuberculosis H37Rv was kindly provided by W. Jacobs, Albert Einstein College of Medicine, Bronx, N.Y.

Aerosol BCG immunization.

Mice were simultaneously infected by using the Glas-col inhalation exposure system (Glas-col Inc., Terre Haute, Ind.) and a log-phase culture of M. bovis BCG in Middlebrook 7H9 broth with an optical density at 600 nm of 0.97. The broth culture was subsequently determined by plating serial dilutions on Middlebrook 7H10 agar to contain 2.3 × 10⁸ CFU/ml.

Intravenous BCG immunization.

Each mouse was injected in the lateral tail vein with 0.2 ml of a 100-fold dilution of the same broth culture used for aerosol immunization; the infectious dose was approximately 5 × 10⁵ CFU/ml.

Aerosol challenge with M. tuberculosis.

Six weeks after immunization, all mice except the nonimmunized and uninfected normal control mice were challenged by the aerosol route by using a broth culture of M. tuberculosis H37Rv containing 5 × 10⁷ CFU/ml in Middlebrook 7H9 broth. The strain had recently been passaged in mice to restore its fitness for mouse infection.

Outcomes.

All mice were weighed weekly. Six mice per group were sacrificed at day 1, week 2, week 4, and week 6 postimmunization and at day 1, week 3, and week 6 postchallenge to determine spleen weights and to obtain quantitative cultures to assess the growth of BCG and/or M. tuberculosis. In addition, at each sacrifice time, the lungs and spleen from one mouse were used to assess the histopathologic changes induced by the immunization strategy.

To monitor the growth of BCG and M. tuberculosis, organ homogenates were plated onto oleate-albumin-dextrose-catalase-enriched Middlebrook 7H10 agar by using appropriate dilutions, and total counts were determined after 4 weeks of incubation at 37°C in the presence of 5% CO₂. Individual BCG and M. tuberculosis counts were determined by plating additional samples on Middlebrook 7H10 agar supplemented with either 30 μg of cycloserine per ml (to which M. bovis BCG Pasteur is naturally resistant) or 50 μg of hygromycin per ml (to which the strain of M. tuberculosis is resistant).

Statistical analysis.

Counts were log transformed before analysis. Pairwise comparisons of group mean values for spleen weights and log₁₀ counts were made by using one-way analysis of variance and Bonferroni's posttest with GraphPad InStat v.3.05 (GraphPad, San Diego, Calif.).

RESULTS

Mouse survival and change in body weight.

All mice, whether they were immunized or not and whether they were challenged or not, survived until they were sacrificed according to protocol. On the day of immunization, the mean body weight of the mice was 16.8 g. The mean body weight increased over the 12 weeks of the study for all groups. There were no significant differences between the groups.

Spleen weight and size.

As soon as 1 day after immunization, the mean spleen weight of intravenously immunized mice was slightly but not significantly higher than that of nonimmunized mice (79 mg versus 66.6 mg). In mice intravenously immunized with BCG, the spleen weight increased to 238 mg by week 4 and then decreased to about 200 mg by week 6 after immunization, and it remained around the same level for the 6 weeks following challenge with M. tuberculosis (Fig. 1). In mice immunized with BCG via the aerosol route, the mean spleen weight began to increase at week 4 and reached the maximum value, 114 mg, by week 6; then it remained at that level following challenge with M. tuberculosis. The spleens of aerosol-immunized mice were significantly smaller than those of intravenously immunized mice at every time examined (P < 0.0001).

FIG. 1.

Mean spleen weights after immunization and after challenge. IV, intravenous; aero, aerosol.

Nonimmunized control mice were also challenged with M. tuberculosis at week 6. The mean spleen weight increased from 70 mg on the day after challenge to 137 and 152 mg after 3 and 6 weeks, respectively. The postchallenge spleen size for nonimmunized but challenged mice was significantly greater than that for nonimmunized unchallenged controls, but it remained less than that for mice immunized intravenously with BCG.

Gross and microscopic lung lesions.

Two weeks after BCG immunization, no gross lesions were visible on the surfaces of lungs from mice immunized by either route. After 4 weeks, gray-white nodules were visible on the lung surfaces in both groups, but they were larger and more numerous in mice immunized by the aerosol route (Fig. 2). After 6 weeks, the size and number of the nodules in both groups had increased, but the nodules were still much larger and more numerous in mice immunized via the aerosol route. On histologic examination, the lungs of mice immunized via the aerosol route harbored many ill-defined peribronchiolar granulomatous lesions, whereas the lungs of intravenously immunized mice harbored fewer lesions that had a predominately perivascular distribution (Fig. 3). Rare acid-fast bacilli were seen in lung sections from both groups.

FIG. 2.

Gross appearance of lungs from nonimmunized mice (a, d, and g), intravenously immunized mice (b, e, and h), and aerosol-immunized mice (c, f, and i) at 4 weeks (a to c) and 6 weeks (d to f) after immunization with BCG and at 6 weeks after challenge with M. tuberculosis (g to i).

FIG. 3.

Histopathology of the lungs of nonimmunized mice (a and d), intravenously immunized mice (b and e), and aerosol-immunized mice (c and f) 6 weeks after BCG immunization (a to c) and 6 weeks after challenge with M. tuberculosis (d to f). Magnification, ×20.

After aerosol challenge with M. tuberculosis, nonimmunized mice developed grossly evident nodular lung lesions that were very small after 3 weeks but much larger and more numerous after 6 weeks (Fig. 2). After 6 weeks, histologic examinations of the lungs demonstrated that there was widespread alveolar consolidation and that there were ill-defined peribronchiolar granulomatous lesions (Fig. 3). In aerosol-immunized mice, the gross lung lesions were more visible on the day after challenge than they were just before challenge, and on histologic examination, the lungs exhibited widespread alveolar filling with cellular exudates, predominantly macrophages, that was much more prominent the day after challenge than it was the day before challenge (Fig. 4). The exudates later cleared in such a way that the gross lung lesions and the histologic appearance of the lungs at 3 and 6 weeks after challenge were similar to the gross lung lesions and the histologic appearance of the lungs observed just before challenge. In mice immunized with BCG intravenously, the gross and histologic lung lesions were similar to the lesions before challenge. By 6 weeks after challenge, the gross and histologic lung lesions were much more limited in mice immunized with BCG by either route than in nonimmunized control mice.

FIG. 4.

Histopathology of the lungs of aerosol-immunized mice 6 weeks after immunization (a and b) and 1 day after challenge with M. tuberculosis (c and d). (a and c) Magnification, ×20; (b and d) magnification, ×200.

Enumeration of BCG in lungs and spleens.

On day 1 after immunization, the lung counts were nearly identical for mice immunized by the intravenous and aerosol routes, 3.65 ± 0.11 and 3.71 ± 0.10 log₁₀CFU, respectively (Table 1). Thereafter, in intravenously immunized mice, the counts increased by 1 log and then remained at a plateau level of 4.4 to 4.8 log₁₀ CFU from week 2 until the end of the experiment. In mice immunized via the aerosol route, the counts reached a peak value of more than 6 log₁₀ CFU by week 4 and then decreased to a plateau level of 4.5 to 4.7 log₁₀ CFU, similar to the value observed in intravenously immunized mice.

TABLE 1.

Multiplication of M. bovis BCG and M. tuberculosis

Organ	Immunization group	Log10 CFU (mean ± SD) after BCG immunization				Log10 CFU (mean ± SD) after aerosol challenge with M. tuberculosis
		1 day	2 wk	4 wk	6 wk	1 day		3 wk		6 wk
						M. tuberculosis	BCG	M. tuberculosis	BCG	M. tuberculosis	BCG
Lung	Nonimmunized					2.37 ± 0.13		6.22 ± 0.05		6.41 ± 0.27
	BCG (intravenous)	3.65 ± 0.11	4.51 ± 0.04	4.74 ± 0.33	4.36 ± 0.28	2.37 ± 0.13	4.36 ± 0.28	4.86 ± 0.29	4.79 ± 0.42	4.76 ± 0.24	4.84 ± 0.23
	BCG (aerosol)	3.71 ± 0.10	4.43 ± 0.38	6.22 ± 0.05	5.97 ± 0.14	2.37 ± 0.13	5.97 ± 0.14	3.87 ± 0.25	4.65 ± 0.11	3.73 ± 0.34	4.49 ± 0.38
Spleen	Nonimmunized					0		3.25 ± 0.93		4.44 ± 0.14
	BCG (intravenous)	4.93 ± 0.07	5.74 ± 0.26	4.81 ± 0.4	5.12 ± 0.23	0	5.12 ± 0.23	0	4.95 ± 0.13	0.86 ± 0.34	5.14 ± 0.13
	BCG (aerosol)	0	0.15 ± 0.26	2.33 ± 0.34	3.54 ± 0.07	0	3.54 ± 0.07	0	3.58 ± 0.45	0	4.16 ± 0.32

In spleens of mice immunized by the intravenous route, the counts were 4.93 ± 0.07 log₁₀ CFU on day 1 after immunization. As in the lungs, the counts increased by about 1 log in the first 2 weeks and reached 5.74 ± 0.26 log₁₀ CFU, and then they remained relatively constant at a plateau level close to 5 log₁₀ CFU until the end of the experiment. In mice immunized via the aerosol route, a limited number of CFU were detected in the spleen of one mouse as early as week 2 after immunization. Thereafter the counts of BCG rapidly increased until the value was 3.5 log₁₀ CFU by week 6 after immunization and then increased more slowly to 4 log₁₀ CFU by 6 weeks after challenge. Clearance of BCG from the lungs or spleens did not occur in either immunization group during the 12-week experiment.

Enumeration of M. tuberculosis after aerosol challenge.

In lungs of nonimmunized controls, the counts of M. tuberculosis were 2.37 ± 0.13 log₁₀ CFU on the day after aerosol challenge (Table 1). After 3 weeks, the counts reached 6.2 log₁₀ CFU and then remained at the same level after 6 weeks. On the other hand, the counts were significantly lower (P < 0.05) in the lungs of mice immunized with BCG by both the aerosol and the intravenous routes than in the lungs of nonimmunized controls at both times. They were also significantly lower (P < 0.05) in aerosol-immunized mice than in intravenously immunized mice. In the former group, the counts remained below 10⁴ CFU/lung, whereas they were 1 log₁₀ higher in the latter group.

In the spleens of nonimmunized controls, the counts were more than 3 log₁₀ CFU at 3 weeks and reached 4.44 ± 0.14 log₁₀ CFU at 6 weeks after challenge. For both BCG-immunized groups of mice, no M. tuberculosis was isolated at 3 weeks. At 6 weeks, 0.86 ± 0.34 log₁₀ CFU was obtained from the spleens of mice immunized by the intravenous route, whereas no M. tuberculosis was isolated from the spleens of mice immunized by the aerosol route.

DISCUSSION

The main finding of the experiments described here is that prior infection with BCG by the aerosol route was highly effective in immunizing mice against subsequent aerosol challenge with virulent M. tuberculosis, resulting in a paucibacillary infection (i.e., <10⁴ organisms per animal). In aerosol-vaccinated mice, multiplication of M. tuberculosis in the lungs was greatly limited after implantation, and no detectable dissemination to the spleen occurred. In addition, as shown previously, prior infection with BCG by the aerosol route had a greater immunizing effect than infection with BCG by the intravenous route had. In the 6 weeks after challenge, the number of tubercle bacilli in the lungs increased by 1, 2, and 4 log₁₀ in aerosol-immunized, intravenously immunized, and nonimmunized mice, respectively. Whereas spleen cultures from aerosol-immunized mice remained free of M. tuberculosis, spleen cultures from intravenously immunized mice did not.

The main objective of the present study was to develop an improved murine model of LTBI for the purpose of evaluating the efficacy of new antimicrobial regimens for use in humans. Only a few models are currently used for this purpose, and each has its disadvantages. The so-called Wayne model is an in vitro model in which a dormant state is induced by exposing a large population of M. tuberculosis, 7 to 8 log₁₀ CFU, to progressive oxygen deprivation (20, 21). Although it is elegant and convenient, this model may not be representative of LTBI in humans, primarily because of the large bacillary population and the anaerobic conditions. The Cornell model (12) is a murine model of microbial persistence induced by antituberculous drugs. Although the procedure presumably results in a paucibacillary state, organisms in a culture-negative state following prolonged antibiotic therapy may be fundamentally different than organisms found in LTBI in humans, which results from the containment of infection by specific host immunity rather than by drugs. In this respect, the previous model of LTBI introduced by Lecoeur and colleagues (9), in which host immunity is augmented by prior immunization with BCG administered intravenously, is more representative of the role of host immunity in the creation of the latent state. However, in this model, as reproduced in the present study, the plateau level of the M. tuberculosis population in the lungs after intravenous BCG immunization is as high as 5 log₁₀ CFU, a level too high to be representative of the bacillary population in a smear-negative lesion.

Therefore, the main findings of the present study provide a method for improving our previous model. Immunization of mice with BCG by the aerosol route and then challenge 6 weeks later with ∼500 CFU of M. tuberculosis result in enhanced immunologic control of the infection, more limited multiplication in the lungs, and a stable paucibacillary state in which the size of the bacillary population in the lungs is less than 4 log₁₀ CFU, characteristics that are more representative of LTBI in humans. Still further reductions in the level at which M. tuberculosis CFU plateau may be possible by using strategies recently shown to immunize mice better than BCG alone, including repeated aerosol BCG immunizations prior to challenge (3) or use of recombinant BCG complemented with a region of deletion-1 (17) or overexpressing antigen 85B (5). Use of a smaller challenge dose of M. tuberculosis may also result in a lower plateau for the bacillary population.

While the main objective of our experiments was to create an improved model of LTBI in humans for the purpose of evaluating new chemotherapeutic strategies, our work has significant overlap with studies aimed at improving the efficacy of the current tuberculosis vaccine strategy that involves subcutaneous BCG injection. The observation that aerosol infection with BCG elicits protection against subsequent aerosol challenge with M. tuberculosis has been made previously in mice (10, 11, 15), as well as in guinea pigs (7, 13) and monkeys (1). In previous studies, aerosol immunization with BCG was consistently shown to be more protective against aerosol challenge than subcutaneous or intradermal immunization (1, 7, 10, 13, 15) and to be more protective against aerosol challenge than intravenous immunization in mice when the two routes were compared directly (10). A second study with mice demonstrated the superior protective efficacy (compared with aerosol challenge) of aerosol immunization over intravenous immunization when the avirulent H37Ra strain of M. tuberculosis was used, but only at the upper range of infective doses used for aerosol immunization (8). Presumably, multiplication of the immunizing strain in the lungs, especially when it is delivered by the aerosol route or the intranasal route, results in a more robust local cellular immune response in the lungs (3) and in the hilar lymph nodes (7), thereby leading to greater containment of a primary lung infection.

More recently, however, two groups of investigators have found that intranasal or aerosol immunization with BCG does not confer greater protection against subsequent aerosol challenge with M. tuberculosis than intravenous and/or subcutaneous immunization confers (3, 16). In each study, there was a delay of at least 18 weeks between immunization and challenge, and BCG was eliminated by the host or by antibiotic treatment prior to challenge. It is possible, therefore, that ongoing infection with the resultant local cellular immune response and inflammation is responsible for the greater protection conferred by aerosol immunization in our study and in previous studies. In the absence of ongoing infection, the route of immunization may have less significance. Palendira et al. have recently shown that aerosol, intravenous, and subcutaneous immunizations result in similar recruitment of gamma interferon-secreting CD4⁺ T cells to the lung and similar protection after aerosol challenge with M. tuberculosis when the BCG used for immunization is eliminated by antibiotic therapy prior to challenge (16).

Finally, BCG vaccination of human subjects by the respiratory route has been reported to be well tolerated and to result in tuberculin conversion in a majority of vaccines (18, 19). While the potential for vaccination of humans by the respiratory route is unclear, it is worth emphasizing that mice immunized and challenged by the aerosol route exhibited an intense cellular response to the challenge that subsided during the 6 weeks after challenge. Aerosol vaccination may therefore confer better protection against aerosol infection at the cost of a stronger delayed-type hypersensitivity response. Such a response may be more harmful in humans who are able to mount a more potent delayed-type hypersensitivity response than mice.