点击显示 收起
Tuberculosis Research Section, Laboratory of Immunogenetics, National Institute for Allergy and Infectious Diseases, National Institutes of Health (NIH), Rockville
Howard Hughes Medical InstituteNIH Research Scholars Program, Bethesda, Maryland
Laboratory of Mycobacterial Immunity and Pathogenesis, Tuberculosis Center, Public Health Research Institute, Newark, New Jersey
Clinical isolates of Mycobacterium tuberculosis demonstrate significant heterogeneity in virulence potential in animal models of infection. Isolate CDC1551, for example, has previously been described in mouse survival studies as being hypovirulent, and isolate HN878 has been described as being hypervirulent. Observed differences in this mouse infection experiment have been proposed to reflect differential engagement of the host immune response. To assess whether this is a local or a systemic effect, C57BL/6 mice were infected simultaneously with mixtures of CDC1551 and HN878 in varying ratios and were monitored for mycobacterial growth kinetics, strain proportions during infection, and mouse survival. Strain mixtures that contained primarily HN878 grew more quickly during the first 5 weeks of infection and were more lethal for mice, and HN878 was enriched during in vivo growth. The absolute number of implanted HN878 bacilli at infection correlated inversely with mouse survival and was independent of concomitant infection with CDC1551. In infections of nonactivated mouse macrophages, HN878 grew more quickly. However, phagocyte preactivation reduced and equalized the growth rate of both strains. These results suggest that HN878 exerts a dominant immunosuppressive effect limited to the granuloma in which it is contained.
Infection with Mycobacterium tuberculosis continues to be a leading cause of death globally. It is estimated that one-third of the world's population is infected with M. tuberculosis, with 8 million new cases of active disease and 2 million fatalities from tuberculosis infection in 1997 [1]. Recent publication of the complete genomic sequence of the laboratory strain H37Rv [2] and clinical isolate CDC1551 [3] has facilitated investigations into the pathogenicity of M. tuberculosis. However, mycobacterial factors that contribute to the virulence of M. tuberculosis or modulate the interaction of this pathogen with the human host are only beginning to be elucidated. Despite previous studies suggesting that M. tuberculosis has an extremely low rate of synonymous mutation [4], there are compelling reasons to suspect that strain-specific attributes contribute directly to virulence and disease outcome. Understanding these attributes would facilitate the pharmacogenomic stratification of patients on the basis of pertinent characteristics of their infecting organism [5].
Clinical isolates of M. tuberculosis have been shown to behave differently in various models of virulence. Strain CDC1551, the causative agent of an outbreak on the Kentucky-Tennessee border between 1994 and 1996, was initially described as hypervirulent in humans, because of high rates of purified protein derivative tuberculin skin-test conversion among those exposed to patients with active disease [6]. However, studies in mouse and rabbit models have suggested that CDC1551 is, in fact, less virulent than other clinical isolates and lab strains, with prolonged survival after infection in mice [7] and slower growth in the rabbit lung [8]. CDC1551 induces relatively high levels of Th1-associated cytokines in infected mouse lungs and in human blood monocytes in vitro. This differential response was shown to be directly attributable to the extractable polar lipid constituents of the cell wall of this clinical isolate [7]. Since Th1 immunity is critical for controlling M. tuberculosis infection, the strong induction of this host response by CDC1551 may serve to protect the host from disease progression. Thus, in the mouse model, this clinical isolate is both hypovirulent and hyperimmunogenic, and we hypothesized that the lipid-associated immunogenicity of CDC1551 may contribute to its attenuated virulence.
In contrast, isolate HN878, the causative agent of an outbreak in Texas between 1995 and 1998 [4], has been shown to cause rapid progression to death in mice in comparison with other clinical isolates and standard lab strains [7, 9]. HN878 belongs to the W-Beijing family of isolates, which have been associated with outbreaks throughout the world and with clusters of drug-resistant disease in the United States [8, 1013]. Multiple studies have indicated that W-Beijing family strains are overrepresented among drug-resistant isolates [1416]; whether the association is merely correlative or a manifestation of unique properties of this family of strains has yet to be determined. HN878 induces a relatively weak Th1-associated cytokine response in the lungs of infected mice [7, 9]. We thus consider HN878 to be both hypervirulent and hypoimmunogenic in mice and hypothesize that its failure to induce a strong protective Th1 response may contribute to its virulence. We have recently shown that the immunosuppressive effect of HN878 was due to the production of a characteristic phenolic glycolipid (PGL-T) by this strain and that deletion of one of the genes encoding its biosynthesis causes it to revert to a less-virulent strain [20]. It is noteworthy that the PGL-Tdeficient HN878 is still not as hypovirulent or hyperimmunogenic as CDC1551 in mouse infections.
For the studies described here, we therefore considered that CDC1551 might produce an immunostimulatory factor that induces a strong Th1 response and thus slows disease progression. Alternatively, or in addition, HN878 might produce an immunosuppressive factor that results in poor induction of immunity and impaired host control of infection. In either scenario, the effects of immunomodulatory products may be either local or systemic. To gain insight into which of these contrasting hypotheses accounts for the observed disparate outcomes of infection, we have infected mice as well as mouse macrophages in vitro with mixed populations of the 2 strains and evaluated bacillary growth kinetics in and survival of the infected mice.
MATERIALS AND METHODS
Growth of strains.
The clinical isolates HN878 and CDC1551 were grown at 37°C in Middlebrook 7H9 medium (Difco) supplemented with 0.05% Tween 80 and albumin-NaCl-glucose complex, to an OD650 of 0.4. Cultures were plated in serial dilutions on 7H11 medium supplemented with oleic acidalbumin-NaCl-glucose complex on days 0, 5, 10, 20, and 30 after infection, for colony enumeration.
Mouse infections.
Established and approved animal-experimentation guidelines were followed in all mouse experiments. Six-week-old C57BL/6 female mice (Taconic) were used for these experiments. Mice were placed into a nose-only apparatus and exposed to an aerosol solution (4 × 106 cfu/mL in the nebulizing chamber) of each strain or strain mixture for 10 min, resulting in the seeding of the lungs with 50 cfu, as determined by harvesting the lungs of 4 mice from each group on the day of infection. Infecting strain mixes were made from frozen glycerol stocks on the basis of the optical density at 650 nm. Lungs of 3 mice from each experimental group were harvested 5 weeks after infection, for colony-forming unit determination. Between 17 and 20 mice in each experimental group were followed for survival time. Mice were killed when they became moribund, and their lungs were harvested for colony-forming unit assay and strain ratio determination.
Bone marrowderived macrophage (BMMP) isolation and infection.
Macrophages were derived from the bone marrow of C57BL/6 mice, as described elsewhere [17]. Adherent cells were suspended in PBS on day 5 and replated at 106 cells/well in 24-well plates. On day 7, cells were infected overnight (1618 h) with M. tuberculosis strains at an MOI of 1 bacillus/10 BMMPs. After infection, cells were washed 3 times with PBS and cultured in Dulbecco's MEM containing 10% fetal bovine serum, 1 mmol/L L-glutamine, and 1% pyruvate (all cell-culture reagents from Hyclone, unless otherwise noted). BMMPs in triplicate wells were lysed in 0.05% SDS on days 0, 3, and 6. Mycobacteria were plated in serial dilutions for colony-forming unit assay and strain ratio determination. For stimulation of BMMPs before infection, mouse interferon (IFN) (R&D Systems) was added to the cells on day 6, at a final concentration of 100 U/mL. After overnight incubation (1518 h), cells were washed 3 times in PBS. Lipopolysaccharide (LPS) (Sigma) was subsequently added, to a final concentration of 10 g/mL, for 34 h. Activated BMMPs were washed 3 times in PBS and subsequently infected.
Strain ratio determination
Individual plated colonies were determined to be CDC1551 or HN878 on the basis of polymerase chain reaction amplification of the genomic region between dnaA and dnaN, which contains an inserted IS6110 element in W-Beijing family M. tuberculosis strains [18]. For strain ratio determination from infected mice, 2030 colonies were typed per mouse. For strain ratio determination from infected BMMPs, at least 30 colonies were typed for each strain mix.
RESULTS
In each experimental group, the proportion of each strain was also determined at the time of death, and HN878 was found to be the predominant strain in every case (figure 1B). Even in mice infected with very high ratios of CDC1551 to HN878 (i.e., 25 : 1), 79% ± 24% (mean ± SE) of the mycobacterial population was composed of HN878 at the stage at which the mice died of infection.
Dominance of the HN878 hyperlethal phenotype in mixed infections of mice.
To assess whether the hypervirulent phenotype of HN878 or the hypovirulent phenotype of CDC1551 dominates in mouse infection with a mixed mycobacterial population, groups of infected mice were monitored for overall survival time. In comparison with mice infected with CDC1551 alone, mice in all other experimental groups progressed to death significantly more rapidly (figure 2A). Even when HN878 made up only 4% of the infecting population, median time to death was significantly shorter than median time to death for mice infected with CDC1551 alone (P = .0036, log rank test).
The lethality of strain HN878 appeared to be strongly dependent on the initial inoculum. The most pronounced virulence phenotype was evident in the experimental group infected with a 1 : 25 mixture of CDC1551 : HN878 rather than in the group infected with only HN878. However, there was significant variability in the average number of colony-forming units implanted between experimental groups (between 25 cfu/lung for HN878 and 107 cfu/lung for the 25 : 1 mixture). When plotted against median time to death for each experimental group, the total number of infecting colony-forming units of strain HN878 was inversely proportional to mouse survival (figure 2B), regardless of the number of CDC1551 colony-forming units in the inoculum. In fact, as little as 2-fold variation in the absolute number of implanted HN878 bacteria had a significant impact on median time to death, suggesting a strong inoculum effect.
Growth rate of CDC1551 and HN878 in in vitro culture.
To assess whether differences in growth rate seen in vivo resulted from inherent differences in mycobacterial strain fitness and growth ability, we studied the growth of CDC1551 and HN878, as well as mixtures of the 2 strains, in ratios of 10 : 1, 1 : 1, and 1 : 10 in liquid culture. All strains and strain mixtures grew at the same rate during both log phase culture (days 05) and entry into stationary phase (days 530) (figure 3). There were no significant differences in doubling times during log phase growth between any of the strains or strain mixtures at any time point (all strains and mixtures had doubling times of 1718 h) (table 1). Thus, differences in growth observed in vivo were not the result of inherent growth characteristics of the 2 strains.
Enhanced growth rate of HN878 late during macrophage infections.
The evidence of an early growth disparity in vivo suggested that the host's innate immune response to infection may play a role in the differential restriction of growth of the 2 strains. To assess the interaction of the 2 strains with infected macrophages, mouse BMMPs were infected with the 2 strains separately or with a mixture of the 2 strains in a 1 : 1 ratio. In unstimulated BMMPs, the 2 strains grew at an equal rate between days 0 and 3 (figure 4A). However, between days 3 and 6, HN878 grew significantly more quickly than CDC1551. This result suggests that, although uptake and initial growth in BMMPs was similar, over time, the interaction of the strains with the cells changed such that the growth rate of CDC1551 was relatively restricted, whereas the growth rate of HN878 was not.
In mixed 1 : 1 infections, the proportion of the 2 strains in BMMP culture remained approximately equal between days 0 and 3 (figure 4B). However, between days 3 and 6, HN878 became the predominant strain (mean ± SE, 70% ± 4%) in the BMMP cultures, suggesting that, although CDC1551 growth had been restricted, HN878 was still replicating in the phagocytes. This effect was observed in 3 independent experiments.
Growth rate of HN878 and CDC1551 in stimulated macrophages.
To address the question of whether disparate strain growth in BMMPs was due to inherent differences in mycobacterial resistance to growth inhibition exerted by the host phagocyte or to differential mycobacterial activation of the infected cells, BMMPs were prestimulated with IFN- and LPS before infection with CDC1551, HN878, and mixtures of the 2 in ratios of 10 : 1, 1 : 1, and 1 : 10. Growth of both strains was significantly slower in preactivated cells than in unstimulated cells. The differences in strain growth observed in unstimulated BMMPs were abolished in prestimulated phagocytes (figure 5A). Thus, HN878 and CDC1551 grew at the same rate during both the early (days 03) and the later (days 36) stages of BMMP infection. In addition, in the 1 : 1 mixture of strains, the infecting ratio did not change significantly between days 3 and 6 (figure 5B). These results suggest that, once the BMMPs were activated by exogenous stimuli, they exerted similar control of growth of HN878 and CDC1551.
DISCUSSION
Although differences in the virulence of clinical isolates of M. tuberculosis have been reported, little is understood about the strain-specific properties that underlie these differences. The observations that we have made in this study about growth kinetics and phenotypic dominance in mixed infections with CDC1551 and HN878 offer insight into the ways that infection with strains of varying virulence results in distinct outcomes.
Our results demonstrate that, within the first 5 weeks of infection, HN878 had a clear growth advantage over CDC1551. The argument could be made that the disparate early growth rates simply reflect metabolic differences between the strains; however, there were no growth differences in liquid culture. Metabolic constraints within infected BMMPs are more likely to reflect the environment of the mouse lung early during infection than in liquid culture, yet the strains did not grow at different rates within the first 3 days of infection of BMMPs. This suggests that the observed difference in growth rates does not reflect an inherent difference in metabolic capability, such as a lack of ability to utilize alternative respiratory pathways or carbon sources. Moreover, CDC1551 does not lack any of the genes of the major metabolic pathways that are expected to play a role during parasitism of host tissues [3]. The faster growth of HN878 relative to that of CDC1551 in infected BMMPs between days 3 and 6, therefore, most likely arises from a differential induction of phagocyte defense mechanisms.
By the time of mouse death, HN878 had become the predominant strain present in all experimental groups infected with mixed populations. Although earlier work indicated that CDC1551 induces a strong Th1 cytokine response observable at the level of whole-lung mRNA [7], in the present study, any immunostimulatory effects were not global enough to protect the host from infection with HN878. Conversely, the presence of HN878 did not facilitate the growth of CDC1551. Thus, either immunomodulation is exerted only in the microenvironment of the organism or the strains have inherent differences in their resistance to killing and growth restriction within macrophages. The fact that differences in growth within BMMPs were abolished when macrophages had been stimulated with IFN- and LPS before infection also supports the idea that the strains do not, in fact, differ in their inherent resistance to growth restriction but, rather, in their induction of macrophage defense mechanisms.
Our mouse survival studies demonstrated that the HN878 hyperlethal phenotype is dominant in mixed infections, with strong penetrance even when HN878 is the minor component of the infecting inoculum. In fact, the simultaneous presence of CDC1551 in infected lungs did not appear to influence the outcome of infection. There was a strong inverse correlation between the number of infecting colony-forming units of HN878 and median time to death, suggestive of an "inoculum effect." In this study, even a 2-fold disparity in the number of implanted colony-forming units had a significant impact on time to death. This result emphasizes the need for caution in interpreting strain-specific differences in mouse survival when the numbers of infecting colony-forming units are not highly similar.
Taken together, our analysis of growth rates in infected lungs and of host survival, as well as the inoculum effect observed for HN878, are consistent with the presence of a dose-dependent immunosuppressive factor such as PGL-T in HN878 [20]. The presence or absence of coinfecting CDC1551 does not appear to significantly alter the course of infection or the outcome, at least in mice. This does not appear to be due to an inherent difference in the ability to withstand macrophage defensesboth strains are comparably growth restricted when macrophages are preactivatedbut, rather, appears to be due to the local lack of activation of macrophages infected with HN878.
Although simultaneous infection with multiple strains of M. tuberculosis may not be prevalent, particularly in areas of low endemicity, recent research in which infecting strains have been genotyped suggests that it does occur [19]. The results of the present study suggest that coinfection may be more difficult to assess than is immediately apparent. Since strain typing is usually performed on isolates recovered from sputum, the results of such tests will not necessarily reflect the mycobacterial population in the whole lung. Our studies suggest that different M. tuberculosis isolates may induce highly localized strain-specific immune responses, whereas bacteria present in the sputum represent only those mycobacteria that have induced sufficient pathologic changes for rupture of the granuloma into airways.
The present study may also have implications for vaccine development. Attenuated strains of M. tuberculosis that are capable of inducing Th1-associated immune responses have long been thought to hold promise as potential vaccine candidates; however, CDC1551, which induces a Th1 cytokine profile, does not protect against simultaneous infection with HN878. Further work would be required to assess whether initial infection with CDC1551 alone could afford better protection against subsequent infection with HN878.
If individual strains or families of strains do, in fact, have clonally expressed immunosuppressive factors, such characteristics could have implications for the outcome of human disease. Rates of treatment failure, relapse, or development of drug resistance may well correlate with properties of the infecting mycobacterial strain. In such a case, we could imagine that duration of treatmentand, perhaps, specific indicated drug regimensmay someday be determined, in part, by identifying the relevant characteristics of the infecting strain.
References
1. Dye C, Scheele S, Dolin P, Pathania V, Raviglione MC. Global burden of tuberculosis: estimated incidence, prevalence, and mortality by country [consensus statement]. WHO Global Surveillance and Monitoring Project. JAMA 1999; 282:67786. First citation in article
2. Cole ST, Brosch R, Parkhill J, et al. Deciphering the biology of Mycobacterium tuberculosis from the complete genome sequence. Nature 1998; 393:53744. First citation in article
3. Fleischmann RD, Alland D, Eisen JA, et al. Whole-genome comparison of Mycobacterium tuberculosis clinical and laboratory strains. J Bacteriol 2002; 184:547990. First citation in article
4. Sreevatsan S, Pan X, Stockbauer KE, et al. Restricted structural gene polymorphism in the Mycobacterium tuberculosis complex indicated evolutionarily recent global dissemination. Proc Natl Acad Sci 1997; 94:986974. First citation in article
5. Kraus CK, Barry CE 3rd. Pharmacogenomics of tuberculosis. In: Rom WN, Garay SM, eds. Tuberculosis. 2nd edition. Philadelphia: Lippincott Williams & Wilkins, 2004:82336. First citation in article
6. Valway SE, Sanchez MP, Shinnick TF, et al. An outbreak involving extensive transmission of a virulent strain of Mycobacterium tuberculosis. N Engl J Med 1998; 338:6339. First citation in article
7. Manca C, Tsenova L, Barry CE III, et al. Mycobacterium tuberculosis CDC1551 induces a more vigorous host response in vivo and in vitro, but is not more virulent than other clinical isolates. J Immunol 1999; 162:67406. First citation in article
8. Tsenova L, Ellison E, Harbacheuski R, et al. Virulence of selected Mycobacterium tuberculosis isolates in the rabbit model of meningitis is dependent on phenolic glycolipid produced by the bacilli. J Infect Dis 2005; 192:98106. First citation in article
9. Manca C, Tsenova L, Bergtold A, et al. Virulence of a Mycobacterium tuberculosis clinical isolate in mice is determined by failure to induce Th1 type immunity and is associated with induction of IFN-/. Proc Natl Acad Sci USA 2001; 98:57527. First citation in article
10. Agerton TB, Valway SE, Blinkhorn RJ, et al. Spread of strain W, a highly drug-resistant strain of Mycobacterium tuberculosis, across the Unites States. Clin Infect Dis 1999; 29:8592. First citation in article
11. Bifani PJ, Mathema B, Liu Z, et al. Identification of a W variant outbreak of Mycobacterium tuberculosis via population-based molecular epidemiology. JAMA 1999; 282:23217. First citation in article
12. Glynn JR, Whitely J, Bifani PJ, Kremer K, van Sooligen D. Worldwide occurrence of Beijing/W strains of Mycobacterium tuberculosis: a systematic review. Emerg Infect Dis 2002; 8:8438. First citation in article
13. Kruuner A, Hoffner SE, Sillastu H, et al. Spread of drug-resistant pulmonary tuberculosis in Estonia. J Clin Microbiol 2001; 39:333944. First citation in article
14. Mokrousov I, Otten T, Vyazovaya A, et al. PCR-based methodology for detecting multidrug-resistant strains of Mycobacterium tuberculosis Beijing family circulating in Russia. Eur J Clin Microbiol Infect Dis 2003; 22:3428. First citation in article
15. Ahn DD, Borgdorff MW, Van LN, et al. Mycobacterium tuberculosis Beijing genotype emerging in Vietnam. Emerg Infect Dis 2000; 6:3025. First citation in article
16. Toungoussova OS, Sandven P, Mariandyshev AO, Nizovtseva NI, Bjune G, Caugant DA. Spread of drug-resistant Mycobacterium tuberculosis strains of the Beijing genotype in the Archangel Oblast, Russia. J Clin Microbiol 2002; 40:19307. First citation in article
17. Scanga CA, Mohan VP, Tanaka K, Alland D, Flynn JL, Chan J. The inducible nitric oxide synthase locus confers protection against aerogenic challenge of both clinical and laboratory strains of Mycobacterium tuberculosis in mice. Infect Immun 2001; 69:77117. First citation in article
18. Kurepina NE, Sreevatsan S, Plikaytis BB, et al. Characterization of the phylogenetic distribution and chromosomal insertion sites of five IS6110 elements in Mycobacterium tuberculosis: non-random integration in the dnaA-dnaN region. Tuber Lung Dis 1998; 79:3142. First citation in article
19. Warren RM, Victor TC, Steicher EM, et al. Patients with active tuberculosis often have different strains in the same sputum specimen. Am J Respir Crit Care Med 2004; 169:6104. First citation in article
20. Reed MB, Domenech P, Manca C, et al. A glycolipid of hypervirulent tuberculosis strains that inhibits the innate immune response. Nature 2004; 431:847. First citation in article