Mycobacterium tuberculosis H37Rv:ΔRD1 Is More Virulent than M. bovis Bacille Calmette-Guérin in Long-Term Murine Infection

Abstract

Region of difference (RD1) genes are present in virulent Mycobacterium tuberculosis but not the vaccine strain M. bovis bacille Calmette-Guérin (BCG). The deletion of RD1 from M. tuberculosis produces an attenuation strikingly like that of BCG, which suggests the use of RD1 mutant strains for improvement of the tuberculosis (TB) vaccine. We performed long-term murine infection with M. tuberculosis H37Rv:ΔRD1 and BCG. Mice infected with H37Rv:ΔRD1 gained less weight than did BCG-infected control mice, and, after >1 year, their lungs harbored many more bacteria and displayed significant levels of inflammation. This difference in virulence has important implications for the pursuit of strains lacking RD1 in the development of the TB vaccine.

Mycobacterium tuberculosis infects nearly 2 billion and kills ∼2 million people annually [1]. In the face of this ongoing global catastrophe, too little is known about the mechanisms of the pathogenesis of M. tuberculosis. A DNA region has been identified that is present in all virulent M. tuberculosis and M. bovis strains but is absent in all strains of the tuberculosis (TB) vaccine M. bovis bacille Calmette-Guérin (BCG) [2-5]. This region of difference 1 (RD1) is required for the virulence of M. tuberculosis. Deletion of the 9.5-kb RD1 region from M. tuberculosis (H37Rv:ΔRD1) results in an attenuation that is strikingly similar to that of BCG in cultured macrophages and mice [6]. By 21 weeks after infection, when lungs from H37Rv-infected mice showed extensive inflammation involving >75% of the lung tissue, the effects of H37Rv:ΔRD1 and BCG remained extremely mild, involving <10% of lung tissue. Lungs from mice infected with H37Rv:ΔRD1 or BCG were indistinguishable from each other. Infection with H37Rv was lethal in these experiments (mean time to death, 32 weeks), whereas all mice infected with H37Rv:ΔRD1 or BCG were alive and apparently healthy [6]. In another study, the introduction of RD1 into BCG resulted in prolonged persistence in immunocompetent mice and greater growth in SCID mice than was seen with the BCG Pasteur parent [7], although virulence was still much reduced relative to M. tuberculosis. The results of recent reports have confirmed the role of RD1 in mycobacterial virulence and have indicated that RD1 genes encode a specialized secretion system, which provides insight into the mechanism by which RD1 exerts its effect [8-11].

Worldwide, >100 million doses of BCG are administered annually [12], but the unceasingly high toll of TB dramatically underscores the need for a better vaccine. The possibility of exploiting RD1 to develop an improved TB vaccine strain is alluring. As a recent derivative of M. tuberculosis, the RD1 mutant probably expresses more relevant antigens than does BCG, which was derived from M. bovis >80 years ago. In addition, the RD1 mutant may persist longer in vivo than BCG, a trait that may be desirable for a TB vaccine. The results of initial animal studies have shown that, in the lung and spleen, H37Rv:ΔRD1 affords protection from challenge very similar to that afforded by BCG and that the restoration of RD1 to BCG might improve the host immune response [8, 9]. To understand better the role of RD1 in the pathogenesis of M. tuberculosis and to help assess its potential to induce long-term protection, we examined mice inoculated with high doses of either the RD1 mutant or BCG. More than 1 year later, we challenged these mice with virulent M. tuberculosis. The results of this challenge were inconclusive because of unexpected differences between the groups of prechallenged mice. Although mice that had been inoculated with BCG appeared to be healthy according to a variety of criteria, the mice infected with the RD1 mutant gained less weight, and, even before the challenge, their lungs had elevated bacteria loads and substantial histopathologic results consistent with TB. Thus, we show here that, although H37Rv:ΔRD1 is greatly attenuated relative to H37Rv, it is, nonetheless, significantly more virulent than BCG.

Materials and methods

M. tuberculosis H37Rv:ΔRD1 and M. bovis BCG-Russia (ATCC 35740) were grown to OD₆₀₀ ∼1.5 in 7H9 medium with 0.05% Tween-80, 0.2% glycerol, and 10% albumin, dextrose, and catalase supplement (BBL) and were stored as 1-mL aliquots in 15% glycerol (final concentration) at −80°C. Aerosol infections of mice were performed as described elsewhere [6]. C57BL/6 mice 6–8 weeks old (Jackson Laboratories) were maintained in a biosafety level 3 animal facility. For infections, frozen bacterial stocks were thawed, sonicated, diluted to ∼10⁶ bacteria/mL, and nebulized in an aerosol infection chamber (Salter Labs or Glas-Col) in which the mice had been placed. The infectious dose was determined by plating whole-lung homogenates from 5 mice in each group to determine the number of colony-forming units on day 1. According to the established protocols of the University of Washington Animal Care Committee, mice were monitored at least twice weekly for signs of disease. Any mouse exhibiting >20% weight loss, trembling, extreme lethargy, or labored breathing was considered to be gravely ill and was killed. The right lung was homogenized in PBS/0.05% Nonidet P-40 and plated as serial dilutions on 7H10 plates. Colonies were counted after 2–3 weeks of incubation at 37°C. Plates with 20–200 colonies were considered to be accurate for the determination of bacterial burden. In some cases, serial dilutions were also plated on 7H10 plates with 30 μg/mL kanamycin, to determine the bacterial burden from the kanamycin-resistant M. tuberculosis challenge strain. In these cases, the bacterial burden from the initial strain was determined by subtracting the number of kanamycin-resistant colonies from the total number of colonies on antibiotic-free plates. To examine histopathologic results, the left lung was removed and inflated, fixed in 10% buffered formalin, embedded in paraffin, sectioned, and stained with hematoxylin-eosin. All histopathologic testing was done in a blinded fashion by at least 3 independent observers.

Results

We have reported previously that, in mice infected with either H37Rv:ΔRD1 or BCG-Russia, bacterial growth and histological results in the lung are strikingly similar to those 21 weeks after infection [6]. To test whether the RD1 mutant harbors any residual virulence and to determine whether it could confer improved long-term protection against fully virulent M. tuberculosis, we initiated longer-term studies. Mice received high doses (∼760 cfu/lung) of either H37Rv:ΔRD1 or BCG. We subsequently infected additional mice with a more moderate dose (∼340 cfu/lung) of either strain. In each case, we infected the mice by use of an aerosol method, to mimic the most common transmission mode of human TB and because M. tuberculosis is more virulent when it is delivered by this route [13].

Infected mice were regularly monitored for signs of TB. Consistent with our earlier findings [6], mice appeared to be free of disease throughout the first year. However, although BCG-infected mice gained weight at a healthy pace, increasing at least 20% during the course of a year, the H37Rv:ΔRD1-infected mice did not gain weight as rapidly (figure 1). This effect was most pronounced in mice that received the higher dose of bacteria (figure 1B).

Figure 1.

Effects of long-term infection with H37:ΔRD1 or bacille Calmette-Guérin (BCG). A and B, Weights of mice infected with a moderate (A, ∼340 cfu) or high (B, ∼760 cfu) dose of BCG-Russia (black circles) or H37:ΔRD1 (white circles). Each point represents the average weight of at least 15 mice. *, Time points beyond which the differences were statistically significant (P<.05). The last weights were recorded immediately before challenge with H37Rv. C, Burdens of H37:ΔRD1 or BCG in lungs of mice after long-term infection. Mice were killed either just before or just after challenge with Mycobacterium tuberculosis. HD, high dose (∼760 cfu); MD, moderate dose (∼340 cfu). Each bar represents the average of 3–5 mice, except at time zero (2 mice/bar). At time zero, the BCG colony-forming units were below the limit of detection (1 × 10⁵ for that time point, as noted by asterisks). RD1, region of difference gene 1.

The failure to gain as much weight suggested that mice infected with H37Rv:ΔRD1 might also show other disease symptoms. Despite this difference, we proceeded to challenge the preinfected mice with a low aerosol dose (5–50 cfu/lung) of virulent M. tuberculosis. The challenge strain H37Rv:MSP12T carried plasmid pMSP12T, which confers resistance to kanamycin, to distinguish it from the preinfecting bacilli. At the time of the challenges, mice that initially received the moderate dose (∼340 cfu/lung) of bacteria had been infected for 57 weeks, whereas mice that initially received the higher dose (∼760 cfu/lung) had been infected for 68 weeks.

Mice were killed at 0, 2, and 7 weeks after challenge, and lung homogenates were plated to determine bacteria numbers. Mice infected with BCG averaged ∼4 × 10⁴ bacteria/lung (figure 1C). However, in contrast to previous findings with infections of shorter duration [6], mice infected long-term with the RD1 mutant harbored 50–500-fold more bacilli per lung at every time point. Comparing all mice analyzed in the experiment, the difference in bacterial burden between those carrying the RD1 mutant and BCG was highly significant (P<.01, unpaired t test).

Mice originally infected with either moderate or high doses of BCG or the RD1 mutant showed roughly equivalent protection against reinfection, with at least 10-fold fewer M. tuberculosis bacilli per lung than in age-matched naive control mice (data not shown). M. tuberculosis bacteria loads were slightly lower in mice preinfected with the RD1 mutant than in those preinfected with BCG, but this difference was not statistically significant. Given the confounding nature of the substantial and unanticipated differences among preinfected mice with regard to weights, bacteria numbers (figure 1), and histopathologic results (figure 2), we chose not to analyze the challenged mice in greater detail.

Figure 2.

C57BL/6 mouse lung tissues after long-term aerosol infection with a high dose (∼760 cfu) of bacille Calmette-Guérin–Russia (A) or H37Rv:ΔRD1 (C) or with a moderate dose (∼340 cfu) of H37Rv:ΔRD1 (B). Times after infection: A and C, 68 weeks; B, 57 weeks. For comparison, lung inflammation produced by H37Rv (infecting dose, ∼100 cfu) after 21 weeks is shown (D). Magnification for each panel, ×40.

Differences in weight and bacteria numbers suggested that mice infected long-term with the RD1 mutant were experiencing active TB. Consistent with these differences, the results of histopathologic analysis of the lungs also revealed marked differences between strains. To highlight potential damage due to the RD1 mutant and to avoid complications arising from reinfection with M. tuberculosis, only mice killed before the M. tuberculosis challenge (time zero in figure 1C) were examined in detail. Lungs from mice infected with BCG were predominantly normal, with very mild lymphocytic infiltration and inflammation restricted to airways, involving ∼5% of the lung tissue (figure 2A). Lungs infected with either the moderate or high dose of BCG were indistinguishable. In contrast, lungs of mice infected with the moderate dose of H37Rv:ΔRD1 displayed significant pathological results that involved ∼50% of the lung tissue (figure 2B). Regions of generally normal lung tissue were interspersed with areas of diffuse, poorly organized inflammation, as well as more focused centers of lymphocytic infiltration that nonetheless lacked the organizational structure characteristic of granulomas. In the lungs of mice infected with the higher dose of H37Rv:ΔRD1 for a longer time, the inflammation was more extensive, involving ∼80% of the lung tissue (figure 2C). Pathological results in these mice were nearly as severe, although not as organized, as that seen in lungs 21 weeks after infection with H37Rv [6] (figure 2D).

Discussion

The global TB pandemic demands a strong response from scientists, clinicians, and public health officials. An efficacious vaccine would be an extremely valuable weapon in efforts to stem the tide of this disease. Among options to achieve this end, researchers are considering new live attenuated bacilli, as well as adding new immunogens to BCG [14]. The identification of RD1 [2] and the subsequent demonstration of its role in the virulence of M. tuberculosis [6] have fueled speculation that this region could be exploited for the development of a vaccine for TB [6, 8, 9]. To assess this possibility and to gain further insight into the pathogenic mechanisms of M. tuberculosis, we performed long-term infection with the RD1 deletion mutant, followed by challenge with wild-type M. tuberculosis.

Here, we have shown that long-term infection with H37Rv:ΔRD1 produces strong indications of disease development. Compared with mice that received BCG, mice infected with H37Rv:ΔRD1 gained less weight (figure 1A and 1B), and, >1 year after infection, their lungs harbored 50–500-fold more bacteria. Most strikingly, the lungs of H37Rv:ΔRD1-infected mice displayed significant symptoms, with both focused lymphocytic infiltrates and areas of diffuse inflammation involving 50% (figure 2B) to 80% (figure 2C) of tissue, whereas BCG-infected lungs remained predominantly normal (figure 2A). Although further analysis is necessary to establish a detailed dose response and time course of pathogenesis, it is notable that prominent disease sequelae were more severe in mice infected with the higher dose of H37Rv:ΔRD1 for a longer time. In fact, just days before M. tuberculosis challenge, 1 mouse infected with the higher dose of H37Rv:ΔRD1 became gravely ill and had to be killed. The lung bacteria load and histopathologic results indicated that this mouse had advanced TB (data not shown). Our findings suggest that other mice infected with the RD1 mutant would also soon have died from TB. In addition, when challenged with M. tuberculosis, mice that harbored the RD1 mutant were not protected significantly better than those that carried BCG. However, considering the failing health of these mice, it seems inappropriate to draw conclusions about the potential role that RD1 genes could play in generating a protective immune response.

Our results contrast sharply with our earlier findings from infections of shorter duration, in which the immune reactions of mice infected with H37Rv:ΔRD1 and BCG were indistinguishable [6]. In SCID mice, the RD1 mutant was somewhat more aggressive than was BCG [7, 9], but these mice are so profoundly immunodeficient that their relevance to the pathogenesis of M. tuberculosis is not wholly clear. The results of a previous study also showed that BALB/c mice infected with a high dose of the RD1 mutant died somewhat earlier than mice infected with BCG [9]. Our demonstration, by a variety of criteria, that the RD1 deletion mutant can cause disease in an immunocompetent mouse that is considered to be highly resistant to M. tuberculosis [15] makes it obvious that there are additional important differences between virulent M. tuberculosis and BCG. In addition, although it is possible to introduce multiple attenuating mutations and to enhance the safety of live vaccines, the residual virulence of the RD1 mutant should serve as a warning that caution is needed when using RD1 or other attenuated mutants for the further development of a TB vaccine.