Deletion of the Mycobacterium tuberculosis Resuscitation-Promoting Factor Rv1009 Gene Results in Delayed Reactivation from Chronic Tuberculo
http://www.100md.com
感染与免疫杂志 2006年第5期
Departments of Medicine Microbiology and Immunology Pathology
The Howard Hughes Medical Institute, Montefiore Medical Center and the Albert Einstein College of Medicine, Bronx, New York 10461
Center for Tuberculosis Research, Department of Medicine, School of Medicine, The Johns Hopkins University, Baltimore, Maryland 21231
ABSTRACT
Approximately one-third of the human population is latently infected with Mycobacterium tuberculosis, comprising a critical reservoir for disease reactivation. Despite the importance of latency in maintaining M. tuberculosis in the human population, little is known about the mycobacterial factors that regulate persistence and reactivation. Previous in vitro studies have implicated a family of five related M. tuberculosis proteins, called resuscitation promoting factors (Rpfs), in regulating mycobacterial growth. We studied the in vivo role of M. tuberculosis rpf genes in an established mouse model of M. tuberculosis persistence and reactivation. After an aerosol infection with the M. tuberculosis Erdman wild type (Erdman) or single-deletion rpf mutants to establish chronic infections in mice, reactivation was induced by administration of the nitric oxide (NO) synthase inhibitor aminoguanidine. Of the five rpf deletion mutants tested, one (Rv1009) exhibited a delayed reactivation phenotype, manifested by delayed postreactivation growth kinetics and prolonged median survival times among infected animals. Immunophenotypic analysis suggested differences in pulmonary B-cell responses between Erdman- and Rv1009-infected mice at advanced stages of reactivation. Analysis of rpf gene expression in the lungs of Erdman-infected mice revealed that relative expression of four of the five rpf-like genes was diminished at late times following reactivation, when bacterial numbers had increased substantially, suggesting that rpf gene expression may be regulated in a growth phase-dependent manner. To our knowledge, Rv1009 is the first M. tuberculosis mutant to have a specific defect in reactivation without accompanying growth defects in vitro or during acute infection in vivo.
INTRODUCTION
Mycobacterium tuberculosis, the causative agent of tuberculosis (TB), is an extraordinarily successful human pathogen. Based on surveys of tuberculin reactivity, it is estimated that up to one-third of the world's population, or 2 billion individuals, is infected with the bacillus. The vast majority of these individuals are asymptomatic latent carriers who exhibit no signs of disease and are noncontagious. However, these latently infected persons represent a critically important reservoir for disease reactivation. For immunocompetent individuals, the risk of reactivation is estimated to be 2 to 23% over their lifetimes; however, for persons coinfected with human immunodeficiency virus, the risk is a considerably higher 10% per year (13, 35). It is crucial that this reservoir of latent infection be addressed if attempts to control the TB epidemic are to succeed.
Although numerous host factors responsible for limiting M. tuberculosis growth have been identified (for a review, see reference 11), much less is known regarding the mycobacterial factors which contribute to persistence, and perhaps even less is known about the mycobacterial signals governing reactivation from a "persistent" or "dormant" state. The resuscitation-promoting factor (Rpf) of Micrococcus luteus, an 16-kDa secreted protein, has been shown to restore active growth to M. luteus cultures rendered dormant due to prolonged incubation in stationary phase (21). The rpf gene homologues of M. tuberculosis, Rv0867c (rpfA), Rv1009 (rpfB), Rv1884c (rpfC), Rv2389c (rpfD), and Rv2450c (rpfE) (http://genolist.pasteur.fr/TubercuList/), have also been demonstrated to stimulate the growth of stationary-phase mycobacteria, including Mycobacterium bovis BCG and M. tuberculosis (24, 45). These studies suggest that the Rpf-like proteins can regulate mycobacterial growth; yet the in vivo function of this gene family remains incompletely defined. The single M. luteus rpf gene is essential for the organism's viability (23); however, deletion of individual rpf genes from the chromosome of M. tuberculosis yielded viable mutants without demonstrable growth defects either in vitro (9, 41) or in vivo in a mouse model of chronic M. tuberculosis infection (10, 41). Microarray analysis of single-deletion rpf mutants during in vitro log-phase growth showed a significant overlap in the expression profiles of several of the mutants, suggesting that they may serve at least partially redundant functions (9) and leading one to speculate that loss of one rpf gene may be compensated by continued expression of the remaining homologues.
The biochemical characterization of Rpf activity has proved to be challenging, due to an apparent lability of in vitro resuscitative effects. This lability may in turn be the result of the reported functional instability of Micrococcus and M. tuberculosis Rpf (24, 37). The solution structure of the Rv1009 core domain, recently solved by heteronuclear multidimensional nuclear magnetic resonance, has been shown to include structural similarity to peptidoglycan-cleaving enzymes of the c-type lysozyme (CAZy family; GH22) and soluble lytic transglycosylase (CAZy family; GH23) families (5). More recently, purified recombinant His-tagged M. luteus Rpf was demonstrated to possess muralytic activity, with the capacity to hydrolyze both fluorescamine-labeled M. luteus cell walls and the synthetic lysozyme substrate 4-methylumbelliferyl--D-N,N',N"-triacetylchitotrioside, while high-level expression of M. luteus Rpf into the Escherichia coli periplasmic space induced bacterial cell lysis (22). Mutational analysis (substitutions of conserved Rpf residues) revealed some correlation between loss of muralytic activity and loss of growth stimulatory activity, although the correlation was imperfect, with some mutations causing substantial impairment of cell wall hydrolytic activity but with minimal effect on culturability (22). It is not immediately evident how such an enzymatic activity would account for the resuscitative effects of the Rpfs on "viable but nonculturable" bacteria and whether the effect is due to a direct alteration of cell wall structure permitting cell growth and division or is mediated more indirectly through the signaling effects of released cell wall components or through altered cell wall permeability.
We and others have exploited genetic systems and animal models to study the rpf gene family. Here, we report the examination of our M. tuberculosis rpf mutants in a murine persistent tuberculosis model. In this model, an initial phase of rapid bacterial growth in the lungs of infected mice is followed by a chronic phase during which tissue bacterial loads remain stable (for a review, see reference 40). Experimental evidence suggests that a static equilibrium between the host and the bacterium is maintained during this chronic phase of infection (25, 30). Thus, in terms of replication dynamics, this chronic persistent infection mimics human latency, although the greater bacterial burden does not reflect what is presumed to be the paucibacillary state of latent TB in humans. We have exploited this model to study the capacity of rpf-deleted M. tuberculosis mutants to reactivate after administration of aminoguanidine (AG), a chemical inhibitor of NO synthase (NOS), which induces disease recrudescence in tuberculous mice by interference with the synthesis of antimycobacterial-reactive nitrogen intermediates (4, 12). We find that the Rv1009 mutant displays delayed kinetics of reactivation, as assessed by both mortality and pulmonary bacterial burden, implicating this rpf-like gene in the regulation of mycobacterial reactivation in vivo.
MATERIALS AND METHODS
Bacterial strains and growth conditions. The wild-type M. tuberculosis Erdman strain (Erdman) (Trudeau Institute, Saranac Lake, NY) and the various rpf homologue-deleted and -complemented strains were grown at 37°C in Middlebrook 7H9 medium supplemented with 0.2% glycerol, 0.05% Tween 80, and 10% oleicacid-albumin-dextrose-catalase enrichment (Becton Dickinson). For growth of the deletion mutants, the medium was supplemented with hygromycin (Roche) at 50 μg/ml, while for growth of the complemented strains the medium was supplemented with both hygromycin at 50 μg/ml and kanamycin (Sigma) at 20 μg/ml.
Construction of rpf-like gene deletion mutants and complemented strains. The construction of the M. tuberculosis strains with deletions of the individual rpf-like genes (Rv0867c, Rv1009, Rv1884c, Rv2389c, or Rv2450c) by the phage-mediated method of specialized transduction has been described previously, as has the construction of the Rv1009 strain complemented at the attB site with Rv1009 (Rv1009 attB::Phsp60 Rv1009) or with Rv1010 (Rv1009 attB::Phsp60 Rv1010) (41). The strain designations with regard to the prior publication have been changed as follows: mc23125 = Rv2389c, mc23126A = Rv1009, mc23127 = Rv2450c, mc23128 = Rv1884c, mc23129 = Rv0867c, mc23301A = Rv1009 attB::Phsp60 Rv1009, and mc23302A = Rv1009 attB::Phsp60 Rv1010 (41).
Mouse infection measures, including organ titers and histopathology. Ten- to 12-week-old C57BL/6 female mice (Charles River) were used in all studies. Animal protocols employed in this study were approved by the Institutional Animal Care and Use Committee. The mice were infected by aerosol (In-Tox Products, Albuquerque, NM) as described previously (32) with various initial inocula (Table 1) of the M. tuberculosis strains. Frozen stocks of the strains were inoculated into 7H9 medium, and cultures were grown to an optical density at 600 nm of 1.0, washed in phosphate-buffered saline (PBS) containing 0.05% Tween 80, and utilized for the mouse infections. Reactivation from the chronic state of infection was induced by the administration of AG (2.5% [wt/vol] ad libitum in drinking water, along with 5% glucose to improve the palatability of the NOS inhibitor), beginning at 16 to 20 weeks postinfection (12). At various times after infection, mice were sacrificed, and portions of the lungs, spleen, and liver were homogenized in PBS with 0.05% Tween 80. Tissue bacterial load was determined by plating dilutions onto Middlebrook 7H10 agar (Difco) supplemented with 0.5% glycerol and 10% oleic acid-albumin-dextrose-catalase. Portions of the various organs were also fixed in 10% phosphate-buffered formalin and embedded in paraffin, and 6-μm sections were subsequently stained with hematoxylin and eosin for histopathological examination (20, 32).
Flow cytometry. To obtain cells from the lungs of M. tuberculosis-infected mice for flow cytometry, pulmonary intravascular blood was flushed via the right ventricle with 10 ml of Dulbecco's modified Eagle medium (DMEM; Invitrogen) containing 2 mM EDTA (GibcoBRL). Lungs were washed in DMEM containing 1-mg/ml heparin and minced with a razor blade in DMEM-heparin. The tissue was then digested with collagenase (1 mg/ml, Type IA Crude; Sigma) and DNase (1,400 IU/ml; Sigma) for 1 h at 37°C. Single-cell suspensions were obtained by passage of the digested material through an 18-gauge needle, followed by passage through a 23-gauge needle and then filtration through a 70-μm nylon strainer (Falcon). Erythrocytes were lysed with ACK lysing buffer (Biosource, Rockville, MD). Cells were then washed once with magnetic cell sorting buffer (PBS supplemented with 2 mM EDTA and 0.5% fetal bovine serum). Treatment with rat anti-mouse CD16/CD32 (FcIII/FcII) Fc blocker (Pharmingen, California) was carried out to prevent nonspecific binding prior to the magnetic labeling of CD45+ cells (Miltenyi Biotec). Magnetic cell sorting separation columns (MS separation columns; Miltenyi Biotec) were used to procure magnetically labeled CD45+ cells according to the manufacturer's protocols. Following magnetic separation, the cells were resuspended in fluorescence-activated cell sorter (FACS) buffer (PBS containing 5% mouse serum and 10% fetal bovine serum) and incubated with fluorochrome-conjugated antibodies. The following fluorochrome-conjugated antibodies were used for this study: anti-CD4-PE, anti-CD8-APC, anti-CD3-PerCP, anti-CD19-PE, anti-LY6G-APC, anti-CD45-APC, and anti-CD45-FITC (all from Pharmingen) and anti-F4/80-PE (from Serotec, Inc.). Following a 40-min incubation with antibodies at 4°C, cells were washed once with FACS buffer, washed with PBS, and fixed overnight at 4°C in 4% paraformaldehyde in PBS. Samples were then subjected to three- or four-color flow cytometric analysis with a FACSCalibur cytometer (Becton Dickinson, San Jose, CA) and FlowJo software (Tree Star Software, San Carlos, Calif.).
Extraction of RNA and analysis of rpf-like gene expression in mouse lungs by real-time PCR. Mouse lungs were homogenized in TRIzol reagent (Invitrogen, Inc.), and mycobacterial RNA was isolated following mechanical disruption with 0.1-mm zirconia-silica beads using a Mini Bead Beater (BioSpec Products), as described previously (26, 41). First-strand cDNA synthesis using Superscript III reverse transcriptase (RT) was carried out according to the manufacturer's instructions, using random hexamers (Roche) as primers and 5 μg of total RNA as a template. Gene-specific PCR primers were designed to amplify 100 to 150 nucleotide fragments with Primer Designer 4 software (Scientific and Educational Software). PCR amplification was carried out using IQ SYBR green supermix (Bio-Rad) according to the following protocol: 95°C for 10 min, followed by 40 cycles, each consisting of 95°C for 30 s, 55°C for 30 s, and 72°C for 30 s. PCR was performed, and data analysis was carried out using the Bio-Rad Laboratories (Hercules, CA) iCyclerIQ real-time PCR detection system. The relative expression of the rpf-like genes following AG-induced reactivation compared with expression in a prereactivation lung sample was computed as previously described (44), with normalization of each rpf-like gene to the expression of 16S rRNA in the same cDNA sample. Non-RT RNA samples were included as negative controls. The primers used were 0867cF (5'-CTCACCGACGGCAATCTGCT-3'), 0867cR (5'-GGCGGCAACTGGTCGATCAA-3'), 1009F (5'-CTGCAGATCTCGCTGGATGG-3'), 1009R (5'-GTGTCGGTCATCGCGAGTTG-3'), 1884cF (5'-TGGATGCCGTGACGAGTCCT-3'), 1884cR (5'-CGGTCGCGGTGCAATAGACA-3'), 2389cF (5'-TTGCCGCCGGATTCGCATTG-3'), 2389cR (5'-TGCGCCAGGATCGTATGCAC-3'), 2450cF (5'-CGGAGCCGGCGGAGTATCG-3'), 2450cR (5'-CCAGCCGGTATCGCCAATG-3'), 16SF (5'-GAGATACTCGAGTGGCGAAC-3'), and 16SR (5'-GGCCGGCTACCCGTCGTC-3').
Statistics. Analysis of survival data was carried out using the Kaplan-Meier method, and the log rank test was used to determine the statistical significance of observed survival differences (GraphPad Prism, version 4.01; GraphPad Software, CA). Where noted, Student's t test was used to determine statistical significance for CFU data.
RESULTS
Rv1009 shows delayed AG-induced reactivation in a mouse model of chronic infection. Mice were infected by aerosol, as described in Materials and Methods, with a relatively high dose (500 to 1,000 CFU) of the M. tuberculosis Erdman wild type or the indicated rpf deletion mutant (Rv0867c, Rv1009, Rv1884c, Rv2389c, or Rv2450c) (see Table 1 for doses). As reported previously (41), all strains showed unimpaired growth in axenic culture, as well as similar initial growth kinetics and capacity to persist in murine tissues, with comparable lung, liver, and spleen bacterial burdens at 4, 8, and 16 weeks postinfection (data not shown).
To induce reactivation from the chronic infection, AG was administered as described in Materials and Methods at 16 weeks postinfection; mice were monitored daily for signs of illness and sacrificed when moribund. Reactivation kinetics after administration of AG are shown in Fig. 1 for "higher-dose" experiments 1A and 1B. For experiment 1A, the time to death upon exposure to AG (expressed as the mean ± standard error of the mean [SEM]) was 63 ± 25 days for Erdman-infected mice and 109 ± 30 days for Rv1009-infected mice. Median times to death were 69 days for Erdman and 112 days for Rv1009 (P < 0.0001 by log rank test). To obtain a mean time to death for the group, the last surviving mouse in the Rv1009 mortality group was recorded as having succumbed at day 168 after AG. This mouse appeared healthy at this point but was sacrificed to determine whether this long-lived survivor was actually infected with M. tuberculosis or whether it escaped infection during the aerosolization process. The mouse was found to have total lung mycobacterial counts of 2.5 x 107, indicating that the prolonged survival occurred despite a significant burden of M. tuberculosis Rv1009 in the tissues. For experiment 1B, the mean survival time (±SEM) after exposure to AG was 68 ± 4 days for Erdman-infected mice and 94 ± 19 days for Rv1009-infected mice, while the median survival time was 77 days for Erdman infection and 97 days for Rv1009. Here, the difference did not achieve statistical significance (P = 0.0649 by log rank test), primarily due to the large variance in the Rv1009-infected group. That is, the first mouse in this study died at 5 days, while the last was sacrificed at 203 days after AG was started (again, as in experiment 1A, the last mouse was sacrificed when well to confirm infection and was found to have >107 bacteria per lung). The very early deaths (within the first week) after AG was administered did not consistently affect any one group, and we suspect that the deaths were unrelated (based upon their timing) to the fatal disease recrudescence with dramatically increased mycobacterial burdens that characterized the later deaths. Although the data we show and the statistical analysis we have performed included all deaths, we note that if one were to censor deaths that occurred within the first week (assuming that they were unrelated to mycobacterial growth resumption based upon their very early timing), then the survival difference in experiment 1B above would reach the level of statistical significance.
The values for initial inocula and for lung bacterial titers just prior to administration of AG (i.e., "prereactivation") are shown in Table 1. For the initial "higher-dose" studies, experiments 1A and 1B, inocula were 2-fold higher for the Rv1009 deletion mutant than for the Erdman wild-type strain. In experiment 2, however, a repetition of the "higher-dose" experiment which also included the complemented strain, the initial inoculum doses were nearly identical, and the finding of delayed reactivation was reproduced. In addition, there was a consistent, strong inverse correlation between the median survival time after infection and the initial inoculum after aerosol infection with many mutant and clinical strains in mouse survival experiments (1, 8). That is, mice receiving the largest initial inocula in a given experiment displayed the shortest survival times. Although in these experiments reactivation was not induced with any specific agent but was allowed to occur through the natural process of senescence and aging, the results would lead us to believe that the approximately twofold-higher inoculum for the Rv1009 strain should actually bias towards a shortened survival time for this group. Therefore, it appears unlikely that the small differences in initial inocula can account for the delayed time to death that we observed for Rv1009-infected mice (Fig. 1A).
In addition, we also note that in experiment 1A the lung titers were about 2.6-fold higher for Erdman-infected than for Rv1009-infected mice at the "prereactivation" time point at 16 weeks postinfection, just prior to administration of AG (Table 1) (P value = 0.015 by Student's t test of log-transformed data). It could be argued that the Rv1009 strain displayed a very subtle persistence defect at these late time points, which accounts in part for the delayed reactivation kinetics which we observe. However, in experiment 1B, as well as in the low-dose experiment 3, there was no statistically significant difference in lung titers during the chronic phase of infection just prior to administration of AG for Erdman- versus Rv1009-infected mice, and yet the pattern of delayed reactivation was reproduced in these studies (P = 0.231 for experiment 1B and P = 0.457 for experiment 3 when log-transformed titers of Erdman and Rv1009 were compared). Other investigators have also found that a Rv1009 deletion mutant (in this case, in the H37Rv background) persists in a manner indistinguishable from the wild type in C57BL/6 mice (10). It is therefore unlikely that these very small differences in lung bacterial burden during the chronic phase of infection, achieving a level of statistical significance in some experiments but not in others, could account for the delay in reactivation that we consistently observed. However, given the trend toward lower mean lung titers during the chronic phase of infection for Rv1009 compared with Erdman, it is possible that the Rv1009 gene has an impact on the capacity of the bacteria to persist at these late stages or perhaps influences the rate of bacterial replication turnover which serves to maintain bacterial numbers during the persistent state.
M. tuberculosis Erdman strains with deletions in each of the four remaining rpf-like genes (Rv0867c, Rv1884c, Rv2389c, and Rv2450c) displayed reactivation kinetics similar to those of the Erdman wild type, as assessed by median survival times (Fig. 1A). After aerosol infection with the Rv0867c mutant, there was a modestly shortened survival time compared with Erdman in the present study (median survival, 53.5 days for Rv0867c-infected mice versus 69 days for Erdman-infected mice), just reaching statistical significance with a P value of 0.0465 on the log rank test. However, because this finding was not reproduced in a separate experiment (not shown), further study with a larger number of mice is necessary to discern whether this accelerated reactivation is a real phenomenon.
Complementation with Rv1009 but not Rv1010 restores reactivation kinetics toward those of the wild type. To further confirm that the delay in postreactivation mortality displayed by Rv1009-infected mice is attributable to loss of Rv1009 expression and not to unintended polar effects on neighboring genes, we complemented the strain by introducing a single copy of Rv1009 at the attB site. Our previous work had shown that disruption of the gene immediately downstream from Rv1009, Rv1010, or ksgA resulted in a small-colony phenotype (41). The Rv1009 mutant we employed for the present studies retained the extreme C terminus of the Rv1009 open reading frame, leaving the overlapping ksgA gene intact; however, this does not ensure that expression of ksgA will be unaffected in the mutant strain. Therefore, for these studies we also constructed a "complemented" strain in which the ksgA gene was introduced into the attB site of the Rv1009 strain. As shown in Fig. 2, reinsertion of Rv1009 into the Rv1009 knockout strain restored the post-AG mortality curve towards that of the wild type (i.e., the median survival time was reduced), although the wild-type phenotype was not completely restored. Median times to death following infection with the indicated strains were 28 days for Erdman, 66 days for Rv1009, and 42 days for the complemented strain Rv1009 attB::Phsp60 Rv1009. Again, as in experiment 1A, the median time to death for Erdman-infected mice was significantly shorter than for their Rv1009-infected counterparts (P = 0.0017 by log rank test). The inability to completely restore the wild-type phenotype by complementation may be due to altered expression of the Rv1009 gene under control of a heterologous promoter (hsp60) or to placement of the complementing gene at the attB site, distant from its native chromosomal locus. In contrast to our findings with Rv1009, introducing a second copy of ksgA (Rv1010) into the Rv1009 strain did not appreciably affect AG-induced mortality (Fig. 2) (median time to death was 69 days for the strain complemented with Rv1010 [Rv1009 attB::Phsp60 Rv1010]), with no significant difference from the parental Rv1009 strain (P = 0.7626 by log rank test). Therefore, it appears that loss of Rv1009 is responsible for the delayed reactivation. Further complementation evidence was provided with the postreactivation organ bacterial loads, as described below (see Fig. 4).
The delayed reactivation phenotype also occurs with low aerosol inoculum of Rv1009. Murine models of M. tuberculosis infection have been described in which the dose of infection has a significant qualitative impact upon the phenotype observed. For instance, both C-C chemokine receptor 2-deficient mice and Toll-like receptor 2-deficient mice display normal control of a low-dose aerogenic infection with M. tuberculosis but are more susceptible than control mice to a high-dose challenge (27, 31, 34). It may be that host defenses that are redundant in one context, and therefore dispensable, can play essential roles in a different experimental setting, such as when antigenic load or bacterial numbers are larger. Given that this altered dependence on specific compensatory mechanisms in different contexts may hold true for the bacillus as well as the host, we decided to examine the growth and reactivation phenotype of the Rv1009 mutant following a low-dose aerosol infection (50 to 100 CFU). The initial (24-h) and prereactivation lung titers are shown in Table 1. As shown in Fig. 3, the Rv1009-infected mice again displayed a delay in reactivation kinetics, as indicated by a delayed time to death after aminoguanidine treatment for the Rv1009 strain. We were unable to calculate median survival times, as the experiment was terminated at day 110 post-AG, with 2 Erdman-infected and 6 Rv1009-infected mice remaining alive (out of a total of 12 mice per group). The organ titers in lung, liver, and spleen were similar at 4 and 18 weeks postinfection (Fig. 4A to C), as we had previously shown for the higher-dose infections (41), again indicating unimpaired in vivo growth kinetics for the mutant.
An early increase in pulmonary bacterial burden and worsened lung histopathology accompanied reactivation in Erdman-infected mice compared with Rv1009-infected mice. Upon administration of AG at 20 weeks postinfection, cohorts of mice were sacrificed at various time intervals to determine organ titers. As shown in Fig. 4A, the Rv1009-infected group had lower lung bacterial burdens after AG administration. After 4 weeks of AG treatment, Erdman lung titers were only 2-fold higher than Rv1009 titers (1.8 x 106 versus 6.8 x 105, respectively). All mice appeared well at this time point. However, by 9 weeks after AG administration, the Erdman-infected mice were appreciably sicker (with ruffled fur, hunched posture, and cachexia), while lung bacterial numbers had dramatically increased and were 1,000-fold higher than lung titers in Rv1009-infected mice. At 11 weeks post-AG, there was a persistent 100-fold difference in bacterial titers (Erdman > Rv1009).
It may be argued that Rv1009-infected mice harbor bacterial titers similar to those seen with Erdman infection, but that the bacteria cannot be cultured on solid medium, due to the rpf defect; that is, they may be "viable but nonculturable" (18). Indeed, it has been reported that lungs harvested from H37Rv-infected mice at 10 months postinfection and plated onto solid medium yielded only about 1.4% as many bacilli as lungs inoculated into liquid culture, indicating that a large proportion of the bacterial population in these chronically infected mice may be unable to grow robustly on plates (7). However, we do not believe that this possibility explains our findings, for two reasons. First, the lung titers did eventually rise in Rv1009-infected mice, as illustrated by the single remaining mouse harvested at 19 weeks post-AG (when no Erdman-infected mice remained alive in this group), which contained 6.5 x 108 CFU/lung by our standard plating techniques. In addition, at the 11-week post-AG harvest, we noted that two mice appeared moribund, one an Erdman-infected mouse and the other a Rv1009-infected mouse. Lungs harvested from each of the two mice were found to harbor 109 bacilli (not shown). This indicates that large bacterial numbers can be recovered from the Rv1009-infected mice by plating bacteria onto solid medium when the mice appear clinically ill.
Complementing the Rv1009 strain with a single copy of Rv1009 at the attB locus restored the postreactivation lung titers to wild-type levels (Fig. 4A), providing evidence that the effect was specific to the loss of Rv1009 expression. Changes in liver titers paralleled those seen in the lungs, with 1,000-fold-higher titers for Erdman-infected mice at 9 weeks post-AG; again, the increase in liver titers occurred at later time points for the Rv1009-infected mice but did eventually reach similar levels. Despite the dramatic differences observed for the pulmonary and hepatic bacterial burdens, the spleen titers rose only slightly (10 fold) for the Erdman-infected mice. The reason for this organ-specific difference in response to AG is unknown but reflects earlier findings examining the kinetics of bacillary proliferation when AG was administered to C57BL/6 mice at 6 months after a low-dose intravenous infection (12). Histopathologic analysis of the lungs 9 weeks after AG administration revealed areas of significant necrosis (Fig. 5A and C; necrotic areas are indicated by black arrows) and progression in the granulomatous response in the lungs of mice infected with both M. tuberculosis Erdman and the Rv1009-complemented strain (Rv1009 attB::Phsp60 Rv1009). In contrast, the Rv1009-infected mice were much less severely affected (Fig. 5B), although there was progression of the granulomatous response after 11 weeks of AG administration (Fig. 5D).
Immunophenotyping reveals increased B lymphocytes in the lungs of M. tuberculosis Erdman-infected mice at late time points postreactivation. The proposed role of Rpfs as putative "bacterial cytokines" has been the primary focus of research on this gene family. It is clearly possible that the 3-log deficit in postreactivation pulmonary bacterial burden in our mouse model was due to loss of a necessary growth-stimulatory function for dormant bacilli. However, it is also possible that the limited and/or delayed postreactivation growth of the Rv1009 null mutant was the result of an altered ability to modulate host responses. For example, loss of Rv1009 may render the bacteria more susceptible to NO-independent host defenses (19). This is admittedly a difficult issue to address, as it is challenging to tease out whether altered host responses precede changes in bacterial growth or whether it is the altered growth which drives differences in host responses.
To begin to address the possibility that the reactivation-deficient phenotype of Rv1009 is due to its ability to elicit a host response different from that triggered by the wild-type Erdman strain, we carried out an immunophenotyping analysis of pulmonary cellular infiltrates at various times after AG treatment. We chose to focus initially on the early (4- to 5-week) post-AG time point, as Erdman and Rv1009 bacterial burdens were very similar and we could eliminate differences in bacterial burden as an additional variable. This immunophenotyping study did not identify any significant quantitative differences in the T cells (CD4+ and CD8+), macrophages (F4/80+), and neutrophils (Ly6G+) in the single-cell suspensions of CD45+ cells obtained from lungs of Erdman- or Rv1009-infected mice (data not shown). These findings corroborate the histopathologic analysis, which revealed no apparent difference at the 4-week time point (not shown). Staining for the CD19 surface marker revealed the absolute number of pulmonic B cells to be 2-fold higher in mice infected with the mutant Rv1009 than in animals challenged with Erdman (Table 2), although this difference did not achieve statistical significance (P = 0.2; Mann-Whitney U test). However, by 10 weeks of AG treatment, the B-cell percentages were markedly higher in the Erdman-infected mice than in their Rv1009-infected counterparts (Fig. 6). The B cells comprised 17.9% ± 6.8% of the total CD45+ lung cell population for the Erdman-infected mice and 5.9% ± 2.0% for the mutant-infected mice (mean ± SEM; three mice per group). As detailed in Table 2, this striking difference in B-cell percentages 10 weeks after AG treatment was attributable primarily to a large influx of B lymphocytes in the Erdman-infected group, while absolute numbers of B cells remained similar for the Rv1009-infected group at the two time points. At 10 weeks of AG treatment, the numbers of B cells per lung were 1.40 x 106 ± 6.1 x 105 for Erdman-infected mice and 2.72 x 105 ± 2.7 x 104 for Rv1009-infected mice (median values were significantly different by the Mann-Whitney U test with a P value of 0.05). Therefore, at this advanced stage of reactivation tuberculosis, there was an increase in both the relative and the absolute numbers of B cells in the lungs of Erdman-infected mice. Because bacterial numbers are dramatically different at this stage of AG administration, we cannot determine whether this represents a direct regulation of B-lymphocyte accumulation by Rv1009 or whether the effect is indirect, due to a delay in the reactivation process and in the resumption of mycobacterial replication.
Expression of M. tuberculosis rpf-like genes after AG-induced reactivation. Although the mycobacterial rpf homologues have been proposed to play roles in the regulation of growth, dormancy, and reactivation, their expression in in vivo models of M. tuberculosis pathogenesis has not been fully explored. It was shown previously, by semiquantitative methods, that transcripts encoding the five rpf family members are expressed in the lungs of mice during the acute phase of infection (2 weeks postinfection) with an M. tuberculosis clinical isolate (41). The present study demonstrates that four of the five M. tuberculosis Erdman rpf homologues (Rv0867c, Rv1009, Rv2389c, and Rv2450c) are also expressed at the RNA level during the chronic persistent stage of infection (Fig. 7). The pulmonary expression level in one "prereactivation" sample (that is, 19 weeks postinfection) was arbitrarily set at a value of 1, and the relative expression in the remaining samples was then evaluated in comparison to this "prereactivation" control sample. To carry out these comparisons, the relative expression of each rpf transcript was normalized to the amount of 16S rRNA, detected by RT-PCR, as an internal control. We found that the relative expression of each of the rpf genes was similar to the "prereactivation" control in the second "prereactivation" mouse, as well in the two mice examined at 4 weeks after AG treatment (Fig. 7), a time point when CFU levels also remained close to prereactivation levels. However, at 9 weeks after AG administration, when pulmonary bacterial burden had escalated to 3 logs above prereactivation levels, the relative expression levels of the four rpf-like genes were found to behave in similar fashions, in that all declined to between 0.1% and 10% of prereactivation levels. This was true for both of the two mice examined, and similarly for the one mouse assessed at the 11-week postreactivation time point (Fig. 7). The Rv2450c gene showed the greatest decrement in relative expression, to only 0.12% of its prereactivation level in the 11-week sample. Although clearly detectable in the 9- and 11-week postreactivation samples (which contained 109 bacilli/lung), the Rv1884c gene could not be reliably detected in the prereactivation lung samples (which contained 106 bacilli/lung); therefore, the relative expression analysis could not be carried out for this gene (data not shown). Overall, the results suggest that rpf gene expression may be regulated in a growth phase-dependent manner in vivo. How this contributes to the role of Rv1009 in reactivation deserves further analysis.
DISCUSSION
Although latency and reactivation are central to the pathogenesis of disease due to M. tuberculosis, many details of the reactivation process remain obscure at the cellular and molecular levels. The M. tuberculosis family of rpf-like genes has been suspected to play a role in regulating reactivation, based in part on in vitro growth-promoting effects of the Rpf-like proteins on stationary-phase bacilli (24). Although control of such a complex process is likely dependent on multiple bacillary factors, as well as numerous host factors (tumor necrosis factor alpha [TNF-], gamma interferon, iNOS, interleukin-12, etc.), we observed a significant effect on the kinetics of reactivation in our mouse model when the Rv1009 gene was deleted from the M. tuberculosis chromosome. To our knowledge, this is the first report of an M. tuberculosis mutant that exhibits unimpaired growth and persistence in a murine model but exhibits a specific defect in the reactivation phase of infection. The reactivation-deficient phenotype of the Rv1009 mutant provides a unique opportunity to characterize host and bacterial responses during reactivation.
In this study, we observed significantly prolonged survival when AG was administered to Rv1009-infected mice in the chronic, persistent phase of infection, compared with M. tuberculosis Erdman wild type-infected controls (Fig. 1A). The extended survival of the Rv1009 group was accompanied by markedly lower pulmonary and hepatic bacterial burdens (Fig. 4) and milder lung immunopathology (Fig. 5), although lung bacterial numbers eventually increased in this group at late stages of infection. The effect occurred with small and large amounts of aerosol inocula, although the survival phenotype was less pronounced during AG-induced reactivation of a low-dose infection (Fig. 3), perhaps due to delayed mortality even for the Erdman-infected mice (2 of 12 mice remained alive at 110 days versus 100% mortality by days 40 to 100 in the various higher-dose experiments), making a further prolongation by the Rv1009 mutant strain more difficult to demonstrate.
These data suggest an in vivo postreactivation growth defect for the Rv1009 bacilli. However, rather than being a growth deficiency intrinsic to the bacillus, it is also possible that this effect is due to altered modulation of host immune responses in mutant-infected mice. Several other M. tuberculosis mutants have been found to modulate host responses and to alter pathogenesis. For example, the pcaA mutant is deficient in -mycolate cyclopropanation, causing deficient cording (14). Purified trehalose dimycolate isolated from the pcaA mutant had altered immunomodulatory activity compared with trehalose dimycolate from wild-type bacteria, inducing significantly lower levels of TNF when applied to murine bone marrow-derived macrophages and provoking a less-potent granulomatous response in vivo (28). This study ties altered M. tuberculosis cell surface structure to a direct alteration in the ability to modulate the host immune response. Similarly, the hypervirulence of M. tuberculosis clinical strain HN878 was linked to production of a unique polyketide synthase-derived phenolic glycolipid, which acts to inhibit the release of proinflammatory mediators in vitro (29). Because Rv1009 is also predicted to be surface expressed and may influence cell wall structure (5), a similar mechanism may be at play in the reactivation-deficient phenotype which we observed. We chose to investigate such a phenotype at an early time (4 to 5 weeks) after AG administration, when bacterial burdens were very similar for the two groups but when Erdman-infected mice were on the verge of developing severe reactivation disease while Rv1009-infected mice remained clinically well for many weeks. Our histopathologic analysis did not reveal differences at this time point. Although immunophenotyping studies (Table 2) revealed increased numbers of B lymphocytes in the lungs of Rv1009-infected mice compared to those infected with the Erdman wild type, the difference did not reach statistical significance. The lack of statistical significance could be due to the fact that, at 4 to 5 weeks postreactivation, the mice infected with wild-type bacilli were in different stages of the reactivation process. This may lead to heterogeneity of the disease state of the mice studied, which in turn results in variability in the immunophenotyping data. Future, more-comprehensive flow cytometric analysis at various time points (early and late) postreactivation and increased sample size may allow more-stringent assessment of the statistical significance of B-lymphocyte observations. An assessment of host responses by microarray and real-time PCR evaluation of host gene expression may also yield further information.
The significance of the immunophenotyping differences which we observed at the later (10-week) postreactivation time points is difficult to interpret, in part due to the confounding variable of the dramatically different bacterial burdens and in part because the importance of B cells in controlling mycobacterial infection remains less than fully defined. B-cell aggregates in the lungs of M. tuberculosis-infected mice (15, 39) and in human tuberculous granulomas (39, 42) have been previously described. However, conflicting data have been reported regarding the importance of B cells in controlling murine M. tuberculosis infection (17, 43). Apart from a role in control of bacterial numbers, B cells have also been implicated in the development of lung pathology (3). Given the ill-defined role for B cells in M. tuberculosis infection, we cannot determine without further study whether the accumulation of B cells in the M. tuberculosis Erdman-infected mice represents a contributing factor to the immunopathology versus an attempted compensatory mechanism to control bacterial numbers and to limit further tissue damage (38). As there has been very limited reporting of immunophenotyping by flow cytometry in animal models of tuberculous reactivation, it would be of interest to explore whether this B-cell accumulation also occurs in other models where reactivation is induced by alternative methods such as steroid administration, CD4 depletion, and TNF- neutralization.
We note that although reactivation was delayed in our model for the Rv1009 mutant, the large expansion of bacillary numbers and development of disease did eventually occur. It is possible that loss of Rv1009 may be compensated by the remaining rpf family members, such that reactivation is slowed but not completely abrogated. In axenic culture, the rpfs have distinct although overlapping expression patterns (41), while microarray analysis of rpf deletion mutants revealed a significant overlap in the global expression profiles during in vitro log-phase growth (9). In addition, M. tuberculosis mutants with single rpf-like gene deletions show growth and persistence similar to those of the wild type in murine infections (10, 41), while M. tuberculosis triple mutants lacking three of the five rpf-like genes have been reported to display a growth defect in a mouse model (1- to 1.5-log-lower lung titers at 16 weeks after intravenous infection) and to show deficits in regrowth in fresh medium after long-term (3.5-month) starvation and oxygen depletion (10). These findings support a redundancy of rpf function.
Our in vivo M. tuberculosis rpf-like gene expression data, obtained during persistence and reactivation (Fig. 7), revealed reduced expression of four of the five rpf homologues at late (9 and 11 weeks) times postreactivation, compared with expression in the chronic persistent phase just prior to the administration of AG. The expression levels vary from 10% of prereactivation levels (for Rv2389c and Rv1009) to 0.1% (for Rv2450c). The results suggest that the various rpf family members may be similarly regulated in a growth phase-dependent manner. The finding that expression of the "resuscitation factors" is actually downregulated during reactivation may be unexpected. However, a full elucidation of the temporal expression of Rv1009 as it relates to M. tuberculosis reactivation in vivo would require an expanded number of time points to more closely examine the changes in expression of Rv1009 which may accompany the switch in bacterial growth state from a condition of nonreplicating persistence to one of active division and expansion in bacterial numbers. That is, it is possible that Rv1009 expression is enhanced as reactivation begins but is downregulated late in the reactivation process (after 9 and 11 weeks of AG treatment) when bacterial numbers have already achieved their peak. It is also possible that additional regulation of Rv1009 occurs at the level of translation or of posttranslational modification; for example, an Rpf family member of Corynebacterium glutamicum has been shown to be glycosylated (16). A further caveat of this type of comparative gene expression study is that the analysis assumes a stable (nonregulated) expression of the internal normalization standard, in our case, the 16S rRNA. Others have found the 16S rRNA copy number per genome to be stable throughout in vitro growth (6). It was also shown that 16S rRNA copy number (measured by real-time RT-PCR) correlates very well with M. tuberculosis CFU numbers in mouse lung (36). However, it has also been shown that the number of ribosomes per cell is increased at higher growth rates (2), raising the possibility that 16S rRNA expression may be regulated in vivo.
It is unknown whether the Rv1009 delayed reactivation phenotype is specific to the use of AG for reactivation and to inhibition of host NO production or whether it reflects a more general phenomenon. Aminoguanidine is known to inhibit NOS but may have additional physiologically relevant effects, such as alteration of polyamine metabolism and modification of ligands for the macrophage scavenger receptor (12). Studying the phenotype of the Rv1009 strain in additional latency models, including CD4 depletion and TNF- neutralization in the context of low-dose M. tuberculosis infections of C57BL/6 mice, may be informative in exploring this issue. The CD4 depletion model is especially attractive, due to its relevance in the setting of human immunodeficiency virus infection and also because in this model the reactivation occurs despite continued expression of gamma interferon and NOS2 (33). This would allow us to determine whether the phenotype is specific to inhibition of macrophage NO production by AG or reflects a more general postreactivation deficit. The TNF- neutralization model (20) is clinically relevant in that TNF blockade therapy for various inflammatory diseases results in reactivation tuberculosis. In addition, this treatment also leads to a decrease in NOS2 expression in lungs at the mRNA level and may therefore share features with the AG reactivation model. Analysis of rpf mutants using these additional reactivation mouse models may further illuminate the mechanisms by which the various M. tuberculosis Rpfs interact with the host in tuberculous infection.
ACKNOWLEDGMENTS
We thank all members of the Chan laboratory for many helpful discussions.
This work was supported by NIH grants AI49375 (J.M.T.), HL71241 (J.C.), and AI26170 (W.R.J.). Generous funding was also provided by the Einstein MMC Center for AIDS Research Developmental Core (NIH NIAID AI51519) and by the Potts Memorial Foundation (J.M.T.).
REFERENCES
1. Barczak, A. K., P. Domenech, H. I. Boshoff, M. B. Reed, C. Manca, G. Kaplan, and C. E. Barry III. 2005. In vivo phenotypic dominance in mouse mixed infections with Mycobacterium tuberculosis clinical isolates. J. Infect. Dis. 192:600-606.
2. Beste, D. J., J. Peters, T. Hooper, C. Avignone-Rossa, M. E. Bushell, and J. McFadden. 2005. Compiling a molecular inventory for Mycobacterium bovis BCG at two growth rates: evidence for growth rate-mediated regulation of ribosome biosynthesis and lipid metabolism. J. Bacteriol. 187:1677-1684.
3. Bosio, C. M., D. Gardner, and K. L. Elkins. 2000. Infection of B cell-deficient mice with CDC 1551, a clinical isolate of Mycobacterium tuberculosis: delay in dissemination and development of lung pathology. J. Immunol. 164:6417-6425.
4. Botha, T., and B. Ryffel. 2002. Reactivation of latent tuberculosis by an inhibitor of inducible nitric oxide synthase in an aerosol murine model. Immunology 107:350-357.
5. Cohen-Gonsaud, M., P. Barthe, C. Bagneris, B. Henderson, J. Ward, C. Roumestand, and N. H. Keep. 2005. The structure of a resuscitation-promoting factor domain from Mycobacterium tuberculosis shows homology to lysozymes. Nat. Struct. Biol. 12:270-273.
6. Desjardin, L. E., L. G. Hayes, C. D. Sohaskey, L. G. Wayne, and K. D. Eisenach. 2001. Microaerophilic induction of the alpha-crystallin chaperone protein homologue (hspX) mRNA of Mycobacterium tuberculosis. J. Bacteriol. 183:5311-5316.
7. Dhillon, J., D. B. Lowrie, and D. A. Mitchison. 2004. Mycobacterium tuberculosis from chronic murine infections that grows in liquid but not on solid medium. BMC Infect. Dis. 4:51.
8. Domenech, P., M. B. Reed, and C. E. Barry III. 2005. Contribution of the Mycobacterium tuberculosis MmpL protein family to virulence and drug resistance. Infect. Immun. 73:3492-3501.
9. Downing, K. J., J. C. Betts, D. I. Young, R. A. McAdam, F. Kelly, M. Young, and V. Mizrahi. 2004. Global expression profiling of strains harbouring null mutations reveals that the five rpf-like genes of Mycobacterium tuberculosis show functional redundancy. Tuberculosis (Edinburgh) 84:167-179.
10. Downing, K. J., V. V. Mischenko, M. O. Shleeva, D. I. Young, M. Young, A. S. Kaprelyants, A. S. Apt, and V. Mizrahi. 2005. Mutants of Mycobacterium tuberculosis lacking three of the five rpf-like genes are defective for growth in vivo and for resuscitation in vitro. Infect. Immun. 73:3038-3043.
11. Flynn, J. L., and J. Chan. 2001. Immunology of tuberculosis. Annu. Rev. Immunol. 19:93-129.
12. Flynn, J. L., C. A. Scanga, K. E. Tanaka, and J. Chan. 1998. Effects of aminoguanidine on latent murine tuberculosis. J. Immunol. 160:1796-1803.
13. Gedde-Dahl, T. 1952. Tuberculous infection in the light of tuberculin matriculation. Am. J. Hyg. 56:139-214.
14. Glickman, M. S., J. S. Cox, and W. R. Jacobs, Jr. 2000. A novel mycolic acid cyclopropane synthetase is required for cording, persistence, and virulence of Mycobacterium tuberculosis. Mol. Cell 5:717-727.
15. Gonzalez-Juarrero, M., O. C. Turner, J. Turner, P. Marietta, J. V. Brooks, and I. M. Orme. 2001. Temporal and spatial arrangement of lymphocytes within lung granulomas induced by aerosol infection with Mycobacterium tuberculosis. Infect. Immun. 69:1722-1728.
16. Hartmann, M., A. Barsch, K. Niehaus, A. Puhler, A. Tauch, and J. Kalinowski. 2004. The glycosylated cell surface protein Rpf2, containing a resuscitation-promoting factor motif, is involved in intercellular communication of Corynebacterium glutamicum. Arch. Microbiol. 182:299-312.
17. Johnson, C. M., A. M. Cooper, A. A. Frank, C. B. Bonorino, L. J. Wysoki, and I. M. Orme. 1997. Mycobacterium tuberculosis aerogenic rechallenge infections in B cell-deficient mice. Tuber. Lung Dis. 78:257-261.
18. Kaprelyants, A. S., J. C. Gottschal, and D. B. Kell. 1993. Dormancy in non-sporulating bacteria. FEMS Microbiol. Rev. 10:271-285.
19. MacMicking, J. D., G. A. Taylor, and J. D. McKinney. 2003. Immune control of tuberculosis by IFN-gamma-inducible LRG-47. Science 302:654-659.
20. Mohan, V. P., C. A. Scanga, K. Yu, H. M. Scott, K. E. Tanaka, E. Tsang, M. C. Tsai, J. L. Flynn, and J. Chan. 2001. Effects of tumor necrosis factor alpha on host immune response in chronic persistent tuberculosis: possible role for limiting pathology. Infect. Immun. 69:1847-1855.
21. Mukamolova, G. V., A. S. Kaprelyants, D. I. Young, M. Young, and D. B. Kell. 1998. A bacterial cytokine. Proc. Natl. Acad. Sci. USA 95:8916-8921.
22. Mukamolova, G. V., A. G. Murzin, E. G. Salina, G. R. Demina, D. B. Kell, A. S. Kaprelyants, and M. Young. 2006. Muralytic activity of Micrococcus luteus Rpf and its relationship to physiological activity in promoting bacterial growth and resuscitation. Mol. Microbiol. 59:84-98.
23. Mukamolova, G. V., O. A. Turapov, K. Kazarian, M. Telkov, A. S. Kaprelyants, D. B. Kell, and M. Young. 2002. The rpf gene of Micrococcus luteus encodes an essential secreted growth factor. Mol. Microbiol. 46:611-621.
24. Mukamolova, G. V., O. A. Turapov, D. I. Young, A. S. Kaprelyants, D. B. Kell, M. Young, K. Kazarian, and M. Telkov. 2002. A family of autocrine growth factors in Mycobacterium tuberculosis. Mol. Microbiol. 46:623-635.
25. Munoz-Elias, E. J., J. Timm, T. Botha, W. T. Chan, J. E. Gomez, and J. D. McKinney. 2005. Replication dynamics of Mycobacterium tuberculosis in chronically infected mice. Infect. Immun. 73:546-551.
26. Ohno, H., G. Zhu, V. P. Mohan, D. Chu, S. Kohno, W. R. Jacobs, Jr., and J. Chan. 2003. The effects of reactive nitrogen intermediates on gene expression in Mycobacterium tuberculosis. Cell. Microbiol. 5:637-648.
27. Peters, W., H. M. Scott, H. F. Chambers, J. L. Flynn, I. F. Charo, and J. D. Ernst. 2001. Chemokine receptor 2 serves an early and essential role in resistance to Mycobacterium tuberculosis. Proc. Natl. Acad. Sci. USA 98:7958-7963.
28. Rao, V., N. Fujiwara, S. A. Porcelli, and M. S. Glickman. 2005. Mycobacterium tuberculosis controls host innate immune activation through cyclopropane modification of a glycolipid effector molecule. J. Exp. Med. 201:535-543.
29. Reed, M. B., P. Domenech, C. Manca, H. Su, A. K. Barczak, B. N. Kreiswirth, G. Kaplan, and C. E. Barry III. 2004. A glycolipid of hypervirulent tuberculosis strains that inhibits the innate immune response. Nature 431:84-87.
30. Rees, R. J., and P. D. Hart. 1961. Analysis of the host-parasite equilibrium in chronic murine tuberculosis by total and viable bacillary counts. Br. J. Exp. Pathol. 42:83-88.
31. Reiling, N., C. Holscher, A. Fehrenbach, S. Kroger, C. J. Kirschning, S. Goyert, and S. Ehlers. 2002. Cutting edge: Toll-like receptor (TLR)2- and TLR4-mediated pathogen recognition in resistance to airborne infection with Mycobacterium tuberculosis. J. Immunol. 169:3480-3484.
32. Scanga, C. A., V. P. Mohan, K. Tanaka, D. Alland, J. L. Flynn, and J. Chan. 2001. The inducible nitric oxide synthase locus confers protection against aerogenic challenge of both clinical and laboratory strains of Mycobacterium tuberculosis in mice. Infect. Immun. 69:7711-7717.
33. Scanga, C. A., V. P. Mohan, K. Yu, H. Joseph, K. Tanaka, J. Chan, and J. L. Flynn. 2000. Depletion of CD4+ T cells causes reactivation of murine persistent tuberculosis despite continued expression of IFN- and nitric oxide synthase 2. J. Exp. Med. 192:347-358.
34. Scott, H. M., and J. L. Flynn. 2002. Mycobacterium tuberculosis in chemokine receptor 2-deficient mice: influence of dose on disease progression. Infect. Immun. 70:5946-5954.
35. Selwyn, P. A., D. Hartel, V. A. Lewis, E. E. Schoenbaum, S. H. Vermund, R. S. Klein, A. T. Walker, and G. H. Friedland. 1989. A prospective study of the risk of tuberculosis among intravenous drug users with human immunodeficiency virus infection. N. Engl. J. Med. 320:545-550.
36. Shi, L., Y. J. Jung, S. Tyagi, M. L. Gennaro, and R. J. North. 2003. Expression of Th1-mediated immunity in mouse lungs induces a Mycobacterium tuberculosis transcription pattern characteristic of nonreplicating persistence. Proc. Natl. Acad. Sci. USA 100:241-246.
37. Shleeva, M. O., K. Bagramyan, M. V. Telkov, G. V. Mukamolova, M. Young, D. B. Kell, and A. S. Kaprelyants. 2002. Formation and resuscitation of "non-culturable" cells of Rhodococcus rhodochrous and Mycobacterium tuberculosis in prolonged stationary phase. Microbiology 148:1581-1591.
38. Taylor, J. L., D. J. Ordway, J. Troudt, M. Gonzalez-Juarrero, R. J. Basaraba, and I. M. Orme. 2005. Factors associated with severe granulomatous pneumonia in Mycobacterium tuberculosis-infected mice vaccinated therapeutically with hsp65 DNA. Infect. Immun. 73:5189-5193.
39. Tsai, M. C., S. Chakravarty, G. Zhu, J. Xu, K. Tanaka, C. Koch, J. Tufariello, J. Flynn, and J. Chan. 2006. Characterization of the tuberculous granuloma in murine and human lungs: cellular composition and relative tissue oxygen tension. Cell. Microbiol. 8:218-232.
40. Tufariello, J. M., J. Chan, and J. L. Flynn. 2003. Latent tuberculosis: mechanisms of host and bacillus that contribute to persistent infection. Lancet Infect. Dis. 3:578-590.
41. Tufariello, J. M., W. R. Jacobs, Jr., and J. Chan. 2004. Individual Mycobacterium tuberculosis resuscitation-promoting factor homologues are dispensable for growth in vitro and in vivo. Infect. Immun. 72:515-526.
42. Ulrichs, T., G. A. Kosmiadi, V. Trusov, S. Jorg, L. Pradl, M. Titukhina, V. Mishenko, N. Gushina, and S. H. Kaufmann. 2004. Human tuberculous granulomas induce peripheral lymphoid follicle-like structures to orchestrate local host defence in the lung. J. Pathol. 204:217-228.
43. Vordermeier, H. M., N. Venkataprasad, D. P. Harris, and J. Ivanyi. 1996. Increase of tuberculous infection in the organs of B cell-deficient mice. Clin. Exp. Immunol. 106:312-316.
44. Zhu, G., H. Xiao, V. P. Mohan, K. Tanaka, S. Tyagi, F. Tsen, P. Salgame, and J. Chan. 2003. Gene expression in the tuberculous granuloma: analysis by laser capture microdissection and real-time PCR. Cell. Microbiol. 5:445-453.
45. Zhu, W., B. B. Plikaytis, and T. M. Shinnick. 2003. Resuscitation factors from mycobacteria: homologs of Micrococcus luteus proteins. Tuberculosis (Edinburgh) 83:261-269.(JoAnn M. Tufariello, Kaix)
The Howard Hughes Medical Institute, Montefiore Medical Center and the Albert Einstein College of Medicine, Bronx, New York 10461
Center for Tuberculosis Research, Department of Medicine, School of Medicine, The Johns Hopkins University, Baltimore, Maryland 21231
ABSTRACT
Approximately one-third of the human population is latently infected with Mycobacterium tuberculosis, comprising a critical reservoir for disease reactivation. Despite the importance of latency in maintaining M. tuberculosis in the human population, little is known about the mycobacterial factors that regulate persistence and reactivation. Previous in vitro studies have implicated a family of five related M. tuberculosis proteins, called resuscitation promoting factors (Rpfs), in regulating mycobacterial growth. We studied the in vivo role of M. tuberculosis rpf genes in an established mouse model of M. tuberculosis persistence and reactivation. After an aerosol infection with the M. tuberculosis Erdman wild type (Erdman) or single-deletion rpf mutants to establish chronic infections in mice, reactivation was induced by administration of the nitric oxide (NO) synthase inhibitor aminoguanidine. Of the five rpf deletion mutants tested, one (Rv1009) exhibited a delayed reactivation phenotype, manifested by delayed postreactivation growth kinetics and prolonged median survival times among infected animals. Immunophenotypic analysis suggested differences in pulmonary B-cell responses between Erdman- and Rv1009-infected mice at advanced stages of reactivation. Analysis of rpf gene expression in the lungs of Erdman-infected mice revealed that relative expression of four of the five rpf-like genes was diminished at late times following reactivation, when bacterial numbers had increased substantially, suggesting that rpf gene expression may be regulated in a growth phase-dependent manner. To our knowledge, Rv1009 is the first M. tuberculosis mutant to have a specific defect in reactivation without accompanying growth defects in vitro or during acute infection in vivo.
INTRODUCTION
Mycobacterium tuberculosis, the causative agent of tuberculosis (TB), is an extraordinarily successful human pathogen. Based on surveys of tuberculin reactivity, it is estimated that up to one-third of the world's population, or 2 billion individuals, is infected with the bacillus. The vast majority of these individuals are asymptomatic latent carriers who exhibit no signs of disease and are noncontagious. However, these latently infected persons represent a critically important reservoir for disease reactivation. For immunocompetent individuals, the risk of reactivation is estimated to be 2 to 23% over their lifetimes; however, for persons coinfected with human immunodeficiency virus, the risk is a considerably higher 10% per year (13, 35). It is crucial that this reservoir of latent infection be addressed if attempts to control the TB epidemic are to succeed.
Although numerous host factors responsible for limiting M. tuberculosis growth have been identified (for a review, see reference 11), much less is known regarding the mycobacterial factors which contribute to persistence, and perhaps even less is known about the mycobacterial signals governing reactivation from a "persistent" or "dormant" state. The resuscitation-promoting factor (Rpf) of Micrococcus luteus, an 16-kDa secreted protein, has been shown to restore active growth to M. luteus cultures rendered dormant due to prolonged incubation in stationary phase (21). The rpf gene homologues of M. tuberculosis, Rv0867c (rpfA), Rv1009 (rpfB), Rv1884c (rpfC), Rv2389c (rpfD), and Rv2450c (rpfE) (http://genolist.pasteur.fr/TubercuList/), have also been demonstrated to stimulate the growth of stationary-phase mycobacteria, including Mycobacterium bovis BCG and M. tuberculosis (24, 45). These studies suggest that the Rpf-like proteins can regulate mycobacterial growth; yet the in vivo function of this gene family remains incompletely defined. The single M. luteus rpf gene is essential for the organism's viability (23); however, deletion of individual rpf genes from the chromosome of M. tuberculosis yielded viable mutants without demonstrable growth defects either in vitro (9, 41) or in vivo in a mouse model of chronic M. tuberculosis infection (10, 41). Microarray analysis of single-deletion rpf mutants during in vitro log-phase growth showed a significant overlap in the expression profiles of several of the mutants, suggesting that they may serve at least partially redundant functions (9) and leading one to speculate that loss of one rpf gene may be compensated by continued expression of the remaining homologues.
The biochemical characterization of Rpf activity has proved to be challenging, due to an apparent lability of in vitro resuscitative effects. This lability may in turn be the result of the reported functional instability of Micrococcus and M. tuberculosis Rpf (24, 37). The solution structure of the Rv1009 core domain, recently solved by heteronuclear multidimensional nuclear magnetic resonance, has been shown to include structural similarity to peptidoglycan-cleaving enzymes of the c-type lysozyme (CAZy family; GH22) and soluble lytic transglycosylase (CAZy family; GH23) families (5). More recently, purified recombinant His-tagged M. luteus Rpf was demonstrated to possess muralytic activity, with the capacity to hydrolyze both fluorescamine-labeled M. luteus cell walls and the synthetic lysozyme substrate 4-methylumbelliferyl--D-N,N',N"-triacetylchitotrioside, while high-level expression of M. luteus Rpf into the Escherichia coli periplasmic space induced bacterial cell lysis (22). Mutational analysis (substitutions of conserved Rpf residues) revealed some correlation between loss of muralytic activity and loss of growth stimulatory activity, although the correlation was imperfect, with some mutations causing substantial impairment of cell wall hydrolytic activity but with minimal effect on culturability (22). It is not immediately evident how such an enzymatic activity would account for the resuscitative effects of the Rpfs on "viable but nonculturable" bacteria and whether the effect is due to a direct alteration of cell wall structure permitting cell growth and division or is mediated more indirectly through the signaling effects of released cell wall components or through altered cell wall permeability.
We and others have exploited genetic systems and animal models to study the rpf gene family. Here, we report the examination of our M. tuberculosis rpf mutants in a murine persistent tuberculosis model. In this model, an initial phase of rapid bacterial growth in the lungs of infected mice is followed by a chronic phase during which tissue bacterial loads remain stable (for a review, see reference 40). Experimental evidence suggests that a static equilibrium between the host and the bacterium is maintained during this chronic phase of infection (25, 30). Thus, in terms of replication dynamics, this chronic persistent infection mimics human latency, although the greater bacterial burden does not reflect what is presumed to be the paucibacillary state of latent TB in humans. We have exploited this model to study the capacity of rpf-deleted M. tuberculosis mutants to reactivate after administration of aminoguanidine (AG), a chemical inhibitor of NO synthase (NOS), which induces disease recrudescence in tuberculous mice by interference with the synthesis of antimycobacterial-reactive nitrogen intermediates (4, 12). We find that the Rv1009 mutant displays delayed kinetics of reactivation, as assessed by both mortality and pulmonary bacterial burden, implicating this rpf-like gene in the regulation of mycobacterial reactivation in vivo.
MATERIALS AND METHODS
Bacterial strains and growth conditions. The wild-type M. tuberculosis Erdman strain (Erdman) (Trudeau Institute, Saranac Lake, NY) and the various rpf homologue-deleted and -complemented strains were grown at 37°C in Middlebrook 7H9 medium supplemented with 0.2% glycerol, 0.05% Tween 80, and 10% oleicacid-albumin-dextrose-catalase enrichment (Becton Dickinson). For growth of the deletion mutants, the medium was supplemented with hygromycin (Roche) at 50 μg/ml, while for growth of the complemented strains the medium was supplemented with both hygromycin at 50 μg/ml and kanamycin (Sigma) at 20 μg/ml.
Construction of rpf-like gene deletion mutants and complemented strains. The construction of the M. tuberculosis strains with deletions of the individual rpf-like genes (Rv0867c, Rv1009, Rv1884c, Rv2389c, or Rv2450c) by the phage-mediated method of specialized transduction has been described previously, as has the construction of the Rv1009 strain complemented at the attB site with Rv1009 (Rv1009 attB::Phsp60 Rv1009) or with Rv1010 (Rv1009 attB::Phsp60 Rv1010) (41). The strain designations with regard to the prior publication have been changed as follows: mc23125 = Rv2389c, mc23126A = Rv1009, mc23127 = Rv2450c, mc23128 = Rv1884c, mc23129 = Rv0867c, mc23301A = Rv1009 attB::Phsp60 Rv1009, and mc23302A = Rv1009 attB::Phsp60 Rv1010 (41).
Mouse infection measures, including organ titers and histopathology. Ten- to 12-week-old C57BL/6 female mice (Charles River) were used in all studies. Animal protocols employed in this study were approved by the Institutional Animal Care and Use Committee. The mice were infected by aerosol (In-Tox Products, Albuquerque, NM) as described previously (32) with various initial inocula (Table 1) of the M. tuberculosis strains. Frozen stocks of the strains were inoculated into 7H9 medium, and cultures were grown to an optical density at 600 nm of 1.0, washed in phosphate-buffered saline (PBS) containing 0.05% Tween 80, and utilized for the mouse infections. Reactivation from the chronic state of infection was induced by the administration of AG (2.5% [wt/vol] ad libitum in drinking water, along with 5% glucose to improve the palatability of the NOS inhibitor), beginning at 16 to 20 weeks postinfection (12). At various times after infection, mice were sacrificed, and portions of the lungs, spleen, and liver were homogenized in PBS with 0.05% Tween 80. Tissue bacterial load was determined by plating dilutions onto Middlebrook 7H10 agar (Difco) supplemented with 0.5% glycerol and 10% oleic acid-albumin-dextrose-catalase. Portions of the various organs were also fixed in 10% phosphate-buffered formalin and embedded in paraffin, and 6-μm sections were subsequently stained with hematoxylin and eosin for histopathological examination (20, 32).
Flow cytometry. To obtain cells from the lungs of M. tuberculosis-infected mice for flow cytometry, pulmonary intravascular blood was flushed via the right ventricle with 10 ml of Dulbecco's modified Eagle medium (DMEM; Invitrogen) containing 2 mM EDTA (GibcoBRL). Lungs were washed in DMEM containing 1-mg/ml heparin and minced with a razor blade in DMEM-heparin. The tissue was then digested with collagenase (1 mg/ml, Type IA Crude; Sigma) and DNase (1,400 IU/ml; Sigma) for 1 h at 37°C. Single-cell suspensions were obtained by passage of the digested material through an 18-gauge needle, followed by passage through a 23-gauge needle and then filtration through a 70-μm nylon strainer (Falcon). Erythrocytes were lysed with ACK lysing buffer (Biosource, Rockville, MD). Cells were then washed once with magnetic cell sorting buffer (PBS supplemented with 2 mM EDTA and 0.5% fetal bovine serum). Treatment with rat anti-mouse CD16/CD32 (FcIII/FcII) Fc blocker (Pharmingen, California) was carried out to prevent nonspecific binding prior to the magnetic labeling of CD45+ cells (Miltenyi Biotec). Magnetic cell sorting separation columns (MS separation columns; Miltenyi Biotec) were used to procure magnetically labeled CD45+ cells according to the manufacturer's protocols. Following magnetic separation, the cells were resuspended in fluorescence-activated cell sorter (FACS) buffer (PBS containing 5% mouse serum and 10% fetal bovine serum) and incubated with fluorochrome-conjugated antibodies. The following fluorochrome-conjugated antibodies were used for this study: anti-CD4-PE, anti-CD8-APC, anti-CD3-PerCP, anti-CD19-PE, anti-LY6G-APC, anti-CD45-APC, and anti-CD45-FITC (all from Pharmingen) and anti-F4/80-PE (from Serotec, Inc.). Following a 40-min incubation with antibodies at 4°C, cells were washed once with FACS buffer, washed with PBS, and fixed overnight at 4°C in 4% paraformaldehyde in PBS. Samples were then subjected to three- or four-color flow cytometric analysis with a FACSCalibur cytometer (Becton Dickinson, San Jose, CA) and FlowJo software (Tree Star Software, San Carlos, Calif.).
Extraction of RNA and analysis of rpf-like gene expression in mouse lungs by real-time PCR. Mouse lungs were homogenized in TRIzol reagent (Invitrogen, Inc.), and mycobacterial RNA was isolated following mechanical disruption with 0.1-mm zirconia-silica beads using a Mini Bead Beater (BioSpec Products), as described previously (26, 41). First-strand cDNA synthesis using Superscript III reverse transcriptase (RT) was carried out according to the manufacturer's instructions, using random hexamers (Roche) as primers and 5 μg of total RNA as a template. Gene-specific PCR primers were designed to amplify 100 to 150 nucleotide fragments with Primer Designer 4 software (Scientific and Educational Software). PCR amplification was carried out using IQ SYBR green supermix (Bio-Rad) according to the following protocol: 95°C for 10 min, followed by 40 cycles, each consisting of 95°C for 30 s, 55°C for 30 s, and 72°C for 30 s. PCR was performed, and data analysis was carried out using the Bio-Rad Laboratories (Hercules, CA) iCyclerIQ real-time PCR detection system. The relative expression of the rpf-like genes following AG-induced reactivation compared with expression in a prereactivation lung sample was computed as previously described (44), with normalization of each rpf-like gene to the expression of 16S rRNA in the same cDNA sample. Non-RT RNA samples were included as negative controls. The primers used were 0867cF (5'-CTCACCGACGGCAATCTGCT-3'), 0867cR (5'-GGCGGCAACTGGTCGATCAA-3'), 1009F (5'-CTGCAGATCTCGCTGGATGG-3'), 1009R (5'-GTGTCGGTCATCGCGAGTTG-3'), 1884cF (5'-TGGATGCCGTGACGAGTCCT-3'), 1884cR (5'-CGGTCGCGGTGCAATAGACA-3'), 2389cF (5'-TTGCCGCCGGATTCGCATTG-3'), 2389cR (5'-TGCGCCAGGATCGTATGCAC-3'), 2450cF (5'-CGGAGCCGGCGGAGTATCG-3'), 2450cR (5'-CCAGCCGGTATCGCCAATG-3'), 16SF (5'-GAGATACTCGAGTGGCGAAC-3'), and 16SR (5'-GGCCGGCTACCCGTCGTC-3').
Statistics. Analysis of survival data was carried out using the Kaplan-Meier method, and the log rank test was used to determine the statistical significance of observed survival differences (GraphPad Prism, version 4.01; GraphPad Software, CA). Where noted, Student's t test was used to determine statistical significance for CFU data.
RESULTS
Rv1009 shows delayed AG-induced reactivation in a mouse model of chronic infection. Mice were infected by aerosol, as described in Materials and Methods, with a relatively high dose (500 to 1,000 CFU) of the M. tuberculosis Erdman wild type or the indicated rpf deletion mutant (Rv0867c, Rv1009, Rv1884c, Rv2389c, or Rv2450c) (see Table 1 for doses). As reported previously (41), all strains showed unimpaired growth in axenic culture, as well as similar initial growth kinetics and capacity to persist in murine tissues, with comparable lung, liver, and spleen bacterial burdens at 4, 8, and 16 weeks postinfection (data not shown).
To induce reactivation from the chronic infection, AG was administered as described in Materials and Methods at 16 weeks postinfection; mice were monitored daily for signs of illness and sacrificed when moribund. Reactivation kinetics after administration of AG are shown in Fig. 1 for "higher-dose" experiments 1A and 1B. For experiment 1A, the time to death upon exposure to AG (expressed as the mean ± standard error of the mean [SEM]) was 63 ± 25 days for Erdman-infected mice and 109 ± 30 days for Rv1009-infected mice. Median times to death were 69 days for Erdman and 112 days for Rv1009 (P < 0.0001 by log rank test). To obtain a mean time to death for the group, the last surviving mouse in the Rv1009 mortality group was recorded as having succumbed at day 168 after AG. This mouse appeared healthy at this point but was sacrificed to determine whether this long-lived survivor was actually infected with M. tuberculosis or whether it escaped infection during the aerosolization process. The mouse was found to have total lung mycobacterial counts of 2.5 x 107, indicating that the prolonged survival occurred despite a significant burden of M. tuberculosis Rv1009 in the tissues. For experiment 1B, the mean survival time (±SEM) after exposure to AG was 68 ± 4 days for Erdman-infected mice and 94 ± 19 days for Rv1009-infected mice, while the median survival time was 77 days for Erdman infection and 97 days for Rv1009. Here, the difference did not achieve statistical significance (P = 0.0649 by log rank test), primarily due to the large variance in the Rv1009-infected group. That is, the first mouse in this study died at 5 days, while the last was sacrificed at 203 days after AG was started (again, as in experiment 1A, the last mouse was sacrificed when well to confirm infection and was found to have >107 bacteria per lung). The very early deaths (within the first week) after AG was administered did not consistently affect any one group, and we suspect that the deaths were unrelated (based upon their timing) to the fatal disease recrudescence with dramatically increased mycobacterial burdens that characterized the later deaths. Although the data we show and the statistical analysis we have performed included all deaths, we note that if one were to censor deaths that occurred within the first week (assuming that they were unrelated to mycobacterial growth resumption based upon their very early timing), then the survival difference in experiment 1B above would reach the level of statistical significance.
The values for initial inocula and for lung bacterial titers just prior to administration of AG (i.e., "prereactivation") are shown in Table 1. For the initial "higher-dose" studies, experiments 1A and 1B, inocula were 2-fold higher for the Rv1009 deletion mutant than for the Erdman wild-type strain. In experiment 2, however, a repetition of the "higher-dose" experiment which also included the complemented strain, the initial inoculum doses were nearly identical, and the finding of delayed reactivation was reproduced. In addition, there was a consistent, strong inverse correlation between the median survival time after infection and the initial inoculum after aerosol infection with many mutant and clinical strains in mouse survival experiments (1, 8). That is, mice receiving the largest initial inocula in a given experiment displayed the shortest survival times. Although in these experiments reactivation was not induced with any specific agent but was allowed to occur through the natural process of senescence and aging, the results would lead us to believe that the approximately twofold-higher inoculum for the Rv1009 strain should actually bias towards a shortened survival time for this group. Therefore, it appears unlikely that the small differences in initial inocula can account for the delayed time to death that we observed for Rv1009-infected mice (Fig. 1A).
In addition, we also note that in experiment 1A the lung titers were about 2.6-fold higher for Erdman-infected than for Rv1009-infected mice at the "prereactivation" time point at 16 weeks postinfection, just prior to administration of AG (Table 1) (P value = 0.015 by Student's t test of log-transformed data). It could be argued that the Rv1009 strain displayed a very subtle persistence defect at these late time points, which accounts in part for the delayed reactivation kinetics which we observe. However, in experiment 1B, as well as in the low-dose experiment 3, there was no statistically significant difference in lung titers during the chronic phase of infection just prior to administration of AG for Erdman- versus Rv1009-infected mice, and yet the pattern of delayed reactivation was reproduced in these studies (P = 0.231 for experiment 1B and P = 0.457 for experiment 3 when log-transformed titers of Erdman and Rv1009 were compared). Other investigators have also found that a Rv1009 deletion mutant (in this case, in the H37Rv background) persists in a manner indistinguishable from the wild type in C57BL/6 mice (10). It is therefore unlikely that these very small differences in lung bacterial burden during the chronic phase of infection, achieving a level of statistical significance in some experiments but not in others, could account for the delay in reactivation that we consistently observed. However, given the trend toward lower mean lung titers during the chronic phase of infection for Rv1009 compared with Erdman, it is possible that the Rv1009 gene has an impact on the capacity of the bacteria to persist at these late stages or perhaps influences the rate of bacterial replication turnover which serves to maintain bacterial numbers during the persistent state.
M. tuberculosis Erdman strains with deletions in each of the four remaining rpf-like genes (Rv0867c, Rv1884c, Rv2389c, and Rv2450c) displayed reactivation kinetics similar to those of the Erdman wild type, as assessed by median survival times (Fig. 1A). After aerosol infection with the Rv0867c mutant, there was a modestly shortened survival time compared with Erdman in the present study (median survival, 53.5 days for Rv0867c-infected mice versus 69 days for Erdman-infected mice), just reaching statistical significance with a P value of 0.0465 on the log rank test. However, because this finding was not reproduced in a separate experiment (not shown), further study with a larger number of mice is necessary to discern whether this accelerated reactivation is a real phenomenon.
Complementation with Rv1009 but not Rv1010 restores reactivation kinetics toward those of the wild type. To further confirm that the delay in postreactivation mortality displayed by Rv1009-infected mice is attributable to loss of Rv1009 expression and not to unintended polar effects on neighboring genes, we complemented the strain by introducing a single copy of Rv1009 at the attB site. Our previous work had shown that disruption of the gene immediately downstream from Rv1009, Rv1010, or ksgA resulted in a small-colony phenotype (41). The Rv1009 mutant we employed for the present studies retained the extreme C terminus of the Rv1009 open reading frame, leaving the overlapping ksgA gene intact; however, this does not ensure that expression of ksgA will be unaffected in the mutant strain. Therefore, for these studies we also constructed a "complemented" strain in which the ksgA gene was introduced into the attB site of the Rv1009 strain. As shown in Fig. 2, reinsertion of Rv1009 into the Rv1009 knockout strain restored the post-AG mortality curve towards that of the wild type (i.e., the median survival time was reduced), although the wild-type phenotype was not completely restored. Median times to death following infection with the indicated strains were 28 days for Erdman, 66 days for Rv1009, and 42 days for the complemented strain Rv1009 attB::Phsp60 Rv1009. Again, as in experiment 1A, the median time to death for Erdman-infected mice was significantly shorter than for their Rv1009-infected counterparts (P = 0.0017 by log rank test). The inability to completely restore the wild-type phenotype by complementation may be due to altered expression of the Rv1009 gene under control of a heterologous promoter (hsp60) or to placement of the complementing gene at the attB site, distant from its native chromosomal locus. In contrast to our findings with Rv1009, introducing a second copy of ksgA (Rv1010) into the Rv1009 strain did not appreciably affect AG-induced mortality (Fig. 2) (median time to death was 69 days for the strain complemented with Rv1010 [Rv1009 attB::Phsp60 Rv1010]), with no significant difference from the parental Rv1009 strain (P = 0.7626 by log rank test). Therefore, it appears that loss of Rv1009 is responsible for the delayed reactivation. Further complementation evidence was provided with the postreactivation organ bacterial loads, as described below (see Fig. 4).
The delayed reactivation phenotype also occurs with low aerosol inoculum of Rv1009. Murine models of M. tuberculosis infection have been described in which the dose of infection has a significant qualitative impact upon the phenotype observed. For instance, both C-C chemokine receptor 2-deficient mice and Toll-like receptor 2-deficient mice display normal control of a low-dose aerogenic infection with M. tuberculosis but are more susceptible than control mice to a high-dose challenge (27, 31, 34). It may be that host defenses that are redundant in one context, and therefore dispensable, can play essential roles in a different experimental setting, such as when antigenic load or bacterial numbers are larger. Given that this altered dependence on specific compensatory mechanisms in different contexts may hold true for the bacillus as well as the host, we decided to examine the growth and reactivation phenotype of the Rv1009 mutant following a low-dose aerosol infection (50 to 100 CFU). The initial (24-h) and prereactivation lung titers are shown in Table 1. As shown in Fig. 3, the Rv1009-infected mice again displayed a delay in reactivation kinetics, as indicated by a delayed time to death after aminoguanidine treatment for the Rv1009 strain. We were unable to calculate median survival times, as the experiment was terminated at day 110 post-AG, with 2 Erdman-infected and 6 Rv1009-infected mice remaining alive (out of a total of 12 mice per group). The organ titers in lung, liver, and spleen were similar at 4 and 18 weeks postinfection (Fig. 4A to C), as we had previously shown for the higher-dose infections (41), again indicating unimpaired in vivo growth kinetics for the mutant.
An early increase in pulmonary bacterial burden and worsened lung histopathology accompanied reactivation in Erdman-infected mice compared with Rv1009-infected mice. Upon administration of AG at 20 weeks postinfection, cohorts of mice were sacrificed at various time intervals to determine organ titers. As shown in Fig. 4A, the Rv1009-infected group had lower lung bacterial burdens after AG administration. After 4 weeks of AG treatment, Erdman lung titers were only 2-fold higher than Rv1009 titers (1.8 x 106 versus 6.8 x 105, respectively). All mice appeared well at this time point. However, by 9 weeks after AG administration, the Erdman-infected mice were appreciably sicker (with ruffled fur, hunched posture, and cachexia), while lung bacterial numbers had dramatically increased and were 1,000-fold higher than lung titers in Rv1009-infected mice. At 11 weeks post-AG, there was a persistent 100-fold difference in bacterial titers (Erdman > Rv1009).
It may be argued that Rv1009-infected mice harbor bacterial titers similar to those seen with Erdman infection, but that the bacteria cannot be cultured on solid medium, due to the rpf defect; that is, they may be "viable but nonculturable" (18). Indeed, it has been reported that lungs harvested from H37Rv-infected mice at 10 months postinfection and plated onto solid medium yielded only about 1.4% as many bacilli as lungs inoculated into liquid culture, indicating that a large proportion of the bacterial population in these chronically infected mice may be unable to grow robustly on plates (7). However, we do not believe that this possibility explains our findings, for two reasons. First, the lung titers did eventually rise in Rv1009-infected mice, as illustrated by the single remaining mouse harvested at 19 weeks post-AG (when no Erdman-infected mice remained alive in this group), which contained 6.5 x 108 CFU/lung by our standard plating techniques. In addition, at the 11-week post-AG harvest, we noted that two mice appeared moribund, one an Erdman-infected mouse and the other a Rv1009-infected mouse. Lungs harvested from each of the two mice were found to harbor 109 bacilli (not shown). This indicates that large bacterial numbers can be recovered from the Rv1009-infected mice by plating bacteria onto solid medium when the mice appear clinically ill.
Complementing the Rv1009 strain with a single copy of Rv1009 at the attB locus restored the postreactivation lung titers to wild-type levels (Fig. 4A), providing evidence that the effect was specific to the loss of Rv1009 expression. Changes in liver titers paralleled those seen in the lungs, with 1,000-fold-higher titers for Erdman-infected mice at 9 weeks post-AG; again, the increase in liver titers occurred at later time points for the Rv1009-infected mice but did eventually reach similar levels. Despite the dramatic differences observed for the pulmonary and hepatic bacterial burdens, the spleen titers rose only slightly (10 fold) for the Erdman-infected mice. The reason for this organ-specific difference in response to AG is unknown but reflects earlier findings examining the kinetics of bacillary proliferation when AG was administered to C57BL/6 mice at 6 months after a low-dose intravenous infection (12). Histopathologic analysis of the lungs 9 weeks after AG administration revealed areas of significant necrosis (Fig. 5A and C; necrotic areas are indicated by black arrows) and progression in the granulomatous response in the lungs of mice infected with both M. tuberculosis Erdman and the Rv1009-complemented strain (Rv1009 attB::Phsp60 Rv1009). In contrast, the Rv1009-infected mice were much less severely affected (Fig. 5B), although there was progression of the granulomatous response after 11 weeks of AG administration (Fig. 5D).
Immunophenotyping reveals increased B lymphocytes in the lungs of M. tuberculosis Erdman-infected mice at late time points postreactivation. The proposed role of Rpfs as putative "bacterial cytokines" has been the primary focus of research on this gene family. It is clearly possible that the 3-log deficit in postreactivation pulmonary bacterial burden in our mouse model was due to loss of a necessary growth-stimulatory function for dormant bacilli. However, it is also possible that the limited and/or delayed postreactivation growth of the Rv1009 null mutant was the result of an altered ability to modulate host responses. For example, loss of Rv1009 may render the bacteria more susceptible to NO-independent host defenses (19). This is admittedly a difficult issue to address, as it is challenging to tease out whether altered host responses precede changes in bacterial growth or whether it is the altered growth which drives differences in host responses.
To begin to address the possibility that the reactivation-deficient phenotype of Rv1009 is due to its ability to elicit a host response different from that triggered by the wild-type Erdman strain, we carried out an immunophenotyping analysis of pulmonary cellular infiltrates at various times after AG treatment. We chose to focus initially on the early (4- to 5-week) post-AG time point, as Erdman and Rv1009 bacterial burdens were very similar and we could eliminate differences in bacterial burden as an additional variable. This immunophenotyping study did not identify any significant quantitative differences in the T cells (CD4+ and CD8+), macrophages (F4/80+), and neutrophils (Ly6G+) in the single-cell suspensions of CD45+ cells obtained from lungs of Erdman- or Rv1009-infected mice (data not shown). These findings corroborate the histopathologic analysis, which revealed no apparent difference at the 4-week time point (not shown). Staining for the CD19 surface marker revealed the absolute number of pulmonic B cells to be 2-fold higher in mice infected with the mutant Rv1009 than in animals challenged with Erdman (Table 2), although this difference did not achieve statistical significance (P = 0.2; Mann-Whitney U test). However, by 10 weeks of AG treatment, the B-cell percentages were markedly higher in the Erdman-infected mice than in their Rv1009-infected counterparts (Fig. 6). The B cells comprised 17.9% ± 6.8% of the total CD45+ lung cell population for the Erdman-infected mice and 5.9% ± 2.0% for the mutant-infected mice (mean ± SEM; three mice per group). As detailed in Table 2, this striking difference in B-cell percentages 10 weeks after AG treatment was attributable primarily to a large influx of B lymphocytes in the Erdman-infected group, while absolute numbers of B cells remained similar for the Rv1009-infected group at the two time points. At 10 weeks of AG treatment, the numbers of B cells per lung were 1.40 x 106 ± 6.1 x 105 for Erdman-infected mice and 2.72 x 105 ± 2.7 x 104 for Rv1009-infected mice (median values were significantly different by the Mann-Whitney U test with a P value of 0.05). Therefore, at this advanced stage of reactivation tuberculosis, there was an increase in both the relative and the absolute numbers of B cells in the lungs of Erdman-infected mice. Because bacterial numbers are dramatically different at this stage of AG administration, we cannot determine whether this represents a direct regulation of B-lymphocyte accumulation by Rv1009 or whether the effect is indirect, due to a delay in the reactivation process and in the resumption of mycobacterial replication.
Expression of M. tuberculosis rpf-like genes after AG-induced reactivation. Although the mycobacterial rpf homologues have been proposed to play roles in the regulation of growth, dormancy, and reactivation, their expression in in vivo models of M. tuberculosis pathogenesis has not been fully explored. It was shown previously, by semiquantitative methods, that transcripts encoding the five rpf family members are expressed in the lungs of mice during the acute phase of infection (2 weeks postinfection) with an M. tuberculosis clinical isolate (41). The present study demonstrates that four of the five M. tuberculosis Erdman rpf homologues (Rv0867c, Rv1009, Rv2389c, and Rv2450c) are also expressed at the RNA level during the chronic persistent stage of infection (Fig. 7). The pulmonary expression level in one "prereactivation" sample (that is, 19 weeks postinfection) was arbitrarily set at a value of 1, and the relative expression in the remaining samples was then evaluated in comparison to this "prereactivation" control sample. To carry out these comparisons, the relative expression of each rpf transcript was normalized to the amount of 16S rRNA, detected by RT-PCR, as an internal control. We found that the relative expression of each of the rpf genes was similar to the "prereactivation" control in the second "prereactivation" mouse, as well in the two mice examined at 4 weeks after AG treatment (Fig. 7), a time point when CFU levels also remained close to prereactivation levels. However, at 9 weeks after AG administration, when pulmonary bacterial burden had escalated to 3 logs above prereactivation levels, the relative expression levels of the four rpf-like genes were found to behave in similar fashions, in that all declined to between 0.1% and 10% of prereactivation levels. This was true for both of the two mice examined, and similarly for the one mouse assessed at the 11-week postreactivation time point (Fig. 7). The Rv2450c gene showed the greatest decrement in relative expression, to only 0.12% of its prereactivation level in the 11-week sample. Although clearly detectable in the 9- and 11-week postreactivation samples (which contained 109 bacilli/lung), the Rv1884c gene could not be reliably detected in the prereactivation lung samples (which contained 106 bacilli/lung); therefore, the relative expression analysis could not be carried out for this gene (data not shown). Overall, the results suggest that rpf gene expression may be regulated in a growth phase-dependent manner in vivo. How this contributes to the role of Rv1009 in reactivation deserves further analysis.
DISCUSSION
Although latency and reactivation are central to the pathogenesis of disease due to M. tuberculosis, many details of the reactivation process remain obscure at the cellular and molecular levels. The M. tuberculosis family of rpf-like genes has been suspected to play a role in regulating reactivation, based in part on in vitro growth-promoting effects of the Rpf-like proteins on stationary-phase bacilli (24). Although control of such a complex process is likely dependent on multiple bacillary factors, as well as numerous host factors (tumor necrosis factor alpha [TNF-], gamma interferon, iNOS, interleukin-12, etc.), we observed a significant effect on the kinetics of reactivation in our mouse model when the Rv1009 gene was deleted from the M. tuberculosis chromosome. To our knowledge, this is the first report of an M. tuberculosis mutant that exhibits unimpaired growth and persistence in a murine model but exhibits a specific defect in the reactivation phase of infection. The reactivation-deficient phenotype of the Rv1009 mutant provides a unique opportunity to characterize host and bacterial responses during reactivation.
In this study, we observed significantly prolonged survival when AG was administered to Rv1009-infected mice in the chronic, persistent phase of infection, compared with M. tuberculosis Erdman wild type-infected controls (Fig. 1A). The extended survival of the Rv1009 group was accompanied by markedly lower pulmonary and hepatic bacterial burdens (Fig. 4) and milder lung immunopathology (Fig. 5), although lung bacterial numbers eventually increased in this group at late stages of infection. The effect occurred with small and large amounts of aerosol inocula, although the survival phenotype was less pronounced during AG-induced reactivation of a low-dose infection (Fig. 3), perhaps due to delayed mortality even for the Erdman-infected mice (2 of 12 mice remained alive at 110 days versus 100% mortality by days 40 to 100 in the various higher-dose experiments), making a further prolongation by the Rv1009 mutant strain more difficult to demonstrate.
These data suggest an in vivo postreactivation growth defect for the Rv1009 bacilli. However, rather than being a growth deficiency intrinsic to the bacillus, it is also possible that this effect is due to altered modulation of host immune responses in mutant-infected mice. Several other M. tuberculosis mutants have been found to modulate host responses and to alter pathogenesis. For example, the pcaA mutant is deficient in -mycolate cyclopropanation, causing deficient cording (14). Purified trehalose dimycolate isolated from the pcaA mutant had altered immunomodulatory activity compared with trehalose dimycolate from wild-type bacteria, inducing significantly lower levels of TNF when applied to murine bone marrow-derived macrophages and provoking a less-potent granulomatous response in vivo (28). This study ties altered M. tuberculosis cell surface structure to a direct alteration in the ability to modulate the host immune response. Similarly, the hypervirulence of M. tuberculosis clinical strain HN878 was linked to production of a unique polyketide synthase-derived phenolic glycolipid, which acts to inhibit the release of proinflammatory mediators in vitro (29). Because Rv1009 is also predicted to be surface expressed and may influence cell wall structure (5), a similar mechanism may be at play in the reactivation-deficient phenotype which we observed. We chose to investigate such a phenotype at an early time (4 to 5 weeks) after AG administration, when bacterial burdens were very similar for the two groups but when Erdman-infected mice were on the verge of developing severe reactivation disease while Rv1009-infected mice remained clinically well for many weeks. Our histopathologic analysis did not reveal differences at this time point. Although immunophenotyping studies (Table 2) revealed increased numbers of B lymphocytes in the lungs of Rv1009-infected mice compared to those infected with the Erdman wild type, the difference did not reach statistical significance. The lack of statistical significance could be due to the fact that, at 4 to 5 weeks postreactivation, the mice infected with wild-type bacilli were in different stages of the reactivation process. This may lead to heterogeneity of the disease state of the mice studied, which in turn results in variability in the immunophenotyping data. Future, more-comprehensive flow cytometric analysis at various time points (early and late) postreactivation and increased sample size may allow more-stringent assessment of the statistical significance of B-lymphocyte observations. An assessment of host responses by microarray and real-time PCR evaluation of host gene expression may also yield further information.
The significance of the immunophenotyping differences which we observed at the later (10-week) postreactivation time points is difficult to interpret, in part due to the confounding variable of the dramatically different bacterial burdens and in part because the importance of B cells in controlling mycobacterial infection remains less than fully defined. B-cell aggregates in the lungs of M. tuberculosis-infected mice (15, 39) and in human tuberculous granulomas (39, 42) have been previously described. However, conflicting data have been reported regarding the importance of B cells in controlling murine M. tuberculosis infection (17, 43). Apart from a role in control of bacterial numbers, B cells have also been implicated in the development of lung pathology (3). Given the ill-defined role for B cells in M. tuberculosis infection, we cannot determine without further study whether the accumulation of B cells in the M. tuberculosis Erdman-infected mice represents a contributing factor to the immunopathology versus an attempted compensatory mechanism to control bacterial numbers and to limit further tissue damage (38). As there has been very limited reporting of immunophenotyping by flow cytometry in animal models of tuberculous reactivation, it would be of interest to explore whether this B-cell accumulation also occurs in other models where reactivation is induced by alternative methods such as steroid administration, CD4 depletion, and TNF- neutralization.
We note that although reactivation was delayed in our model for the Rv1009 mutant, the large expansion of bacillary numbers and development of disease did eventually occur. It is possible that loss of Rv1009 may be compensated by the remaining rpf family members, such that reactivation is slowed but not completely abrogated. In axenic culture, the rpfs have distinct although overlapping expression patterns (41), while microarray analysis of rpf deletion mutants revealed a significant overlap in the global expression profiles during in vitro log-phase growth (9). In addition, M. tuberculosis mutants with single rpf-like gene deletions show growth and persistence similar to those of the wild type in murine infections (10, 41), while M. tuberculosis triple mutants lacking three of the five rpf-like genes have been reported to display a growth defect in a mouse model (1- to 1.5-log-lower lung titers at 16 weeks after intravenous infection) and to show deficits in regrowth in fresh medium after long-term (3.5-month) starvation and oxygen depletion (10). These findings support a redundancy of rpf function.
Our in vivo M. tuberculosis rpf-like gene expression data, obtained during persistence and reactivation (Fig. 7), revealed reduced expression of four of the five rpf homologues at late (9 and 11 weeks) times postreactivation, compared with expression in the chronic persistent phase just prior to the administration of AG. The expression levels vary from 10% of prereactivation levels (for Rv2389c and Rv1009) to 0.1% (for Rv2450c). The results suggest that the various rpf family members may be similarly regulated in a growth phase-dependent manner. The finding that expression of the "resuscitation factors" is actually downregulated during reactivation may be unexpected. However, a full elucidation of the temporal expression of Rv1009 as it relates to M. tuberculosis reactivation in vivo would require an expanded number of time points to more closely examine the changes in expression of Rv1009 which may accompany the switch in bacterial growth state from a condition of nonreplicating persistence to one of active division and expansion in bacterial numbers. That is, it is possible that Rv1009 expression is enhanced as reactivation begins but is downregulated late in the reactivation process (after 9 and 11 weeks of AG treatment) when bacterial numbers have already achieved their peak. It is also possible that additional regulation of Rv1009 occurs at the level of translation or of posttranslational modification; for example, an Rpf family member of Corynebacterium glutamicum has been shown to be glycosylated (16). A further caveat of this type of comparative gene expression study is that the analysis assumes a stable (nonregulated) expression of the internal normalization standard, in our case, the 16S rRNA. Others have found the 16S rRNA copy number per genome to be stable throughout in vitro growth (6). It was also shown that 16S rRNA copy number (measured by real-time RT-PCR) correlates very well with M. tuberculosis CFU numbers in mouse lung (36). However, it has also been shown that the number of ribosomes per cell is increased at higher growth rates (2), raising the possibility that 16S rRNA expression may be regulated in vivo.
It is unknown whether the Rv1009 delayed reactivation phenotype is specific to the use of AG for reactivation and to inhibition of host NO production or whether it reflects a more general phenomenon. Aminoguanidine is known to inhibit NOS but may have additional physiologically relevant effects, such as alteration of polyamine metabolism and modification of ligands for the macrophage scavenger receptor (12). Studying the phenotype of the Rv1009 strain in additional latency models, including CD4 depletion and TNF- neutralization in the context of low-dose M. tuberculosis infections of C57BL/6 mice, may be informative in exploring this issue. The CD4 depletion model is especially attractive, due to its relevance in the setting of human immunodeficiency virus infection and also because in this model the reactivation occurs despite continued expression of gamma interferon and NOS2 (33). This would allow us to determine whether the phenotype is specific to inhibition of macrophage NO production by AG or reflects a more general postreactivation deficit. The TNF- neutralization model (20) is clinically relevant in that TNF blockade therapy for various inflammatory diseases results in reactivation tuberculosis. In addition, this treatment also leads to a decrease in NOS2 expression in lungs at the mRNA level and may therefore share features with the AG reactivation model. Analysis of rpf mutants using these additional reactivation mouse models may further illuminate the mechanisms by which the various M. tuberculosis Rpfs interact with the host in tuberculous infection.
ACKNOWLEDGMENTS
We thank all members of the Chan laboratory for many helpful discussions.
This work was supported by NIH grants AI49375 (J.M.T.), HL71241 (J.C.), and AI26170 (W.R.J.). Generous funding was also provided by the Einstein MMC Center for AIDS Research Developmental Core (NIH NIAID AI51519) and by the Potts Memorial Foundation (J.M.T.).
REFERENCES
1. Barczak, A. K., P. Domenech, H. I. Boshoff, M. B. Reed, C. Manca, G. Kaplan, and C. E. Barry III. 2005. In vivo phenotypic dominance in mouse mixed infections with Mycobacterium tuberculosis clinical isolates. J. Infect. Dis. 192:600-606.
2. Beste, D. J., J. Peters, T. Hooper, C. Avignone-Rossa, M. E. Bushell, and J. McFadden. 2005. Compiling a molecular inventory for Mycobacterium bovis BCG at two growth rates: evidence for growth rate-mediated regulation of ribosome biosynthesis and lipid metabolism. J. Bacteriol. 187:1677-1684.
3. Bosio, C. M., D. Gardner, and K. L. Elkins. 2000. Infection of B cell-deficient mice with CDC 1551, a clinical isolate of Mycobacterium tuberculosis: delay in dissemination and development of lung pathology. J. Immunol. 164:6417-6425.
4. Botha, T., and B. Ryffel. 2002. Reactivation of latent tuberculosis by an inhibitor of inducible nitric oxide synthase in an aerosol murine model. Immunology 107:350-357.
5. Cohen-Gonsaud, M., P. Barthe, C. Bagneris, B. Henderson, J. Ward, C. Roumestand, and N. H. Keep. 2005. The structure of a resuscitation-promoting factor domain from Mycobacterium tuberculosis shows homology to lysozymes. Nat. Struct. Biol. 12:270-273.
6. Desjardin, L. E., L. G. Hayes, C. D. Sohaskey, L. G. Wayne, and K. D. Eisenach. 2001. Microaerophilic induction of the alpha-crystallin chaperone protein homologue (hspX) mRNA of Mycobacterium tuberculosis. J. Bacteriol. 183:5311-5316.
7. Dhillon, J., D. B. Lowrie, and D. A. Mitchison. 2004. Mycobacterium tuberculosis from chronic murine infections that grows in liquid but not on solid medium. BMC Infect. Dis. 4:51.
8. Domenech, P., M. B. Reed, and C. E. Barry III. 2005. Contribution of the Mycobacterium tuberculosis MmpL protein family to virulence and drug resistance. Infect. Immun. 73:3492-3501.
9. Downing, K. J., J. C. Betts, D. I. Young, R. A. McAdam, F. Kelly, M. Young, and V. Mizrahi. 2004. Global expression profiling of strains harbouring null mutations reveals that the five rpf-like genes of Mycobacterium tuberculosis show functional redundancy. Tuberculosis (Edinburgh) 84:167-179.
10. Downing, K. J., V. V. Mischenko, M. O. Shleeva, D. I. Young, M. Young, A. S. Kaprelyants, A. S. Apt, and V. Mizrahi. 2005. Mutants of Mycobacterium tuberculosis lacking three of the five rpf-like genes are defective for growth in vivo and for resuscitation in vitro. Infect. Immun. 73:3038-3043.
11. Flynn, J. L., and J. Chan. 2001. Immunology of tuberculosis. Annu. Rev. Immunol. 19:93-129.
12. Flynn, J. L., C. A. Scanga, K. E. Tanaka, and J. Chan. 1998. Effects of aminoguanidine on latent murine tuberculosis. J. Immunol. 160:1796-1803.
13. Gedde-Dahl, T. 1952. Tuberculous infection in the light of tuberculin matriculation. Am. J. Hyg. 56:139-214.
14. Glickman, M. S., J. S. Cox, and W. R. Jacobs, Jr. 2000. A novel mycolic acid cyclopropane synthetase is required for cording, persistence, and virulence of Mycobacterium tuberculosis. Mol. Cell 5:717-727.
15. Gonzalez-Juarrero, M., O. C. Turner, J. Turner, P. Marietta, J. V. Brooks, and I. M. Orme. 2001. Temporal and spatial arrangement of lymphocytes within lung granulomas induced by aerosol infection with Mycobacterium tuberculosis. Infect. Immun. 69:1722-1728.
16. Hartmann, M., A. Barsch, K. Niehaus, A. Puhler, A. Tauch, and J. Kalinowski. 2004. The glycosylated cell surface protein Rpf2, containing a resuscitation-promoting factor motif, is involved in intercellular communication of Corynebacterium glutamicum. Arch. Microbiol. 182:299-312.
17. Johnson, C. M., A. M. Cooper, A. A. Frank, C. B. Bonorino, L. J. Wysoki, and I. M. Orme. 1997. Mycobacterium tuberculosis aerogenic rechallenge infections in B cell-deficient mice. Tuber. Lung Dis. 78:257-261.
18. Kaprelyants, A. S., J. C. Gottschal, and D. B. Kell. 1993. Dormancy in non-sporulating bacteria. FEMS Microbiol. Rev. 10:271-285.
19. MacMicking, J. D., G. A. Taylor, and J. D. McKinney. 2003. Immune control of tuberculosis by IFN-gamma-inducible LRG-47. Science 302:654-659.
20. Mohan, V. P., C. A. Scanga, K. Yu, H. M. Scott, K. E. Tanaka, E. Tsang, M. C. Tsai, J. L. Flynn, and J. Chan. 2001. Effects of tumor necrosis factor alpha on host immune response in chronic persistent tuberculosis: possible role for limiting pathology. Infect. Immun. 69:1847-1855.
21. Mukamolova, G. V., A. S. Kaprelyants, D. I. Young, M. Young, and D. B. Kell. 1998. A bacterial cytokine. Proc. Natl. Acad. Sci. USA 95:8916-8921.
22. Mukamolova, G. V., A. G. Murzin, E. G. Salina, G. R. Demina, D. B. Kell, A. S. Kaprelyants, and M. Young. 2006. Muralytic activity of Micrococcus luteus Rpf and its relationship to physiological activity in promoting bacterial growth and resuscitation. Mol. Microbiol. 59:84-98.
23. Mukamolova, G. V., O. A. Turapov, K. Kazarian, M. Telkov, A. S. Kaprelyants, D. B. Kell, and M. Young. 2002. The rpf gene of Micrococcus luteus encodes an essential secreted growth factor. Mol. Microbiol. 46:611-621.
24. Mukamolova, G. V., O. A. Turapov, D. I. Young, A. S. Kaprelyants, D. B. Kell, M. Young, K. Kazarian, and M. Telkov. 2002. A family of autocrine growth factors in Mycobacterium tuberculosis. Mol. Microbiol. 46:623-635.
25. Munoz-Elias, E. J., J. Timm, T. Botha, W. T. Chan, J. E. Gomez, and J. D. McKinney. 2005. Replication dynamics of Mycobacterium tuberculosis in chronically infected mice. Infect. Immun. 73:546-551.
26. Ohno, H., G. Zhu, V. P. Mohan, D. Chu, S. Kohno, W. R. Jacobs, Jr., and J. Chan. 2003. The effects of reactive nitrogen intermediates on gene expression in Mycobacterium tuberculosis. Cell. Microbiol. 5:637-648.
27. Peters, W., H. M. Scott, H. F. Chambers, J. L. Flynn, I. F. Charo, and J. D. Ernst. 2001. Chemokine receptor 2 serves an early and essential role in resistance to Mycobacterium tuberculosis. Proc. Natl. Acad. Sci. USA 98:7958-7963.
28. Rao, V., N. Fujiwara, S. A. Porcelli, and M. S. Glickman. 2005. Mycobacterium tuberculosis controls host innate immune activation through cyclopropane modification of a glycolipid effector molecule. J. Exp. Med. 201:535-543.
29. Reed, M. B., P. Domenech, C. Manca, H. Su, A. K. Barczak, B. N. Kreiswirth, G. Kaplan, and C. E. Barry III. 2004. A glycolipid of hypervirulent tuberculosis strains that inhibits the innate immune response. Nature 431:84-87.
30. Rees, R. J., and P. D. Hart. 1961. Analysis of the host-parasite equilibrium in chronic murine tuberculosis by total and viable bacillary counts. Br. J. Exp. Pathol. 42:83-88.
31. Reiling, N., C. Holscher, A. Fehrenbach, S. Kroger, C. J. Kirschning, S. Goyert, and S. Ehlers. 2002. Cutting edge: Toll-like receptor (TLR)2- and TLR4-mediated pathogen recognition in resistance to airborne infection with Mycobacterium tuberculosis. J. Immunol. 169:3480-3484.
32. Scanga, C. A., V. P. Mohan, K. Tanaka, D. Alland, J. L. Flynn, and J. Chan. 2001. The inducible nitric oxide synthase locus confers protection against aerogenic challenge of both clinical and laboratory strains of Mycobacterium tuberculosis in mice. Infect. Immun. 69:7711-7717.
33. Scanga, C. A., V. P. Mohan, K. Yu, H. Joseph, K. Tanaka, J. Chan, and J. L. Flynn. 2000. Depletion of CD4+ T cells causes reactivation of murine persistent tuberculosis despite continued expression of IFN- and nitric oxide synthase 2. J. Exp. Med. 192:347-358.
34. Scott, H. M., and J. L. Flynn. 2002. Mycobacterium tuberculosis in chemokine receptor 2-deficient mice: influence of dose on disease progression. Infect. Immun. 70:5946-5954.
35. Selwyn, P. A., D. Hartel, V. A. Lewis, E. E. Schoenbaum, S. H. Vermund, R. S. Klein, A. T. Walker, and G. H. Friedland. 1989. A prospective study of the risk of tuberculosis among intravenous drug users with human immunodeficiency virus infection. N. Engl. J. Med. 320:545-550.
36. Shi, L., Y. J. Jung, S. Tyagi, M. L. Gennaro, and R. J. North. 2003. Expression of Th1-mediated immunity in mouse lungs induces a Mycobacterium tuberculosis transcription pattern characteristic of nonreplicating persistence. Proc. Natl. Acad. Sci. USA 100:241-246.
37. Shleeva, M. O., K. Bagramyan, M. V. Telkov, G. V. Mukamolova, M. Young, D. B. Kell, and A. S. Kaprelyants. 2002. Formation and resuscitation of "non-culturable" cells of Rhodococcus rhodochrous and Mycobacterium tuberculosis in prolonged stationary phase. Microbiology 148:1581-1591.
38. Taylor, J. L., D. J. Ordway, J. Troudt, M. Gonzalez-Juarrero, R. J. Basaraba, and I. M. Orme. 2005. Factors associated with severe granulomatous pneumonia in Mycobacterium tuberculosis-infected mice vaccinated therapeutically with hsp65 DNA. Infect. Immun. 73:5189-5193.
39. Tsai, M. C., S. Chakravarty, G. Zhu, J. Xu, K. Tanaka, C. Koch, J. Tufariello, J. Flynn, and J. Chan. 2006. Characterization of the tuberculous granuloma in murine and human lungs: cellular composition and relative tissue oxygen tension. Cell. Microbiol. 8:218-232.
40. Tufariello, J. M., J. Chan, and J. L. Flynn. 2003. Latent tuberculosis: mechanisms of host and bacillus that contribute to persistent infection. Lancet Infect. Dis. 3:578-590.
41. Tufariello, J. M., W. R. Jacobs, Jr., and J. Chan. 2004. Individual Mycobacterium tuberculosis resuscitation-promoting factor homologues are dispensable for growth in vitro and in vivo. Infect. Immun. 72:515-526.
42. Ulrichs, T., G. A. Kosmiadi, V. Trusov, S. Jorg, L. Pradl, M. Titukhina, V. Mishenko, N. Gushina, and S. H. Kaufmann. 2004. Human tuberculous granulomas induce peripheral lymphoid follicle-like structures to orchestrate local host defence in the lung. J. Pathol. 204:217-228.
43. Vordermeier, H. M., N. Venkataprasad, D. P. Harris, and J. Ivanyi. 1996. Increase of tuberculous infection in the organs of B cell-deficient mice. Clin. Exp. Immunol. 106:312-316.
44. Zhu, G., H. Xiao, V. P. Mohan, K. Tanaka, S. Tyagi, F. Tsen, P. Salgame, and J. Chan. 2003. Gene expression in the tuberculous granuloma: analysis by laser capture microdissection and real-time PCR. Cell. Microbiol. 5:445-453.
45. Zhu, W., B. B. Plikaytis, and T. M. Shinnick. 2003. Resuscitation factors from mycobacteria: homologs of Micrococcus luteus proteins. Tuberculosis (Edinburgh) 83:261-269.(JoAnn M. Tufariello, Kaix)