Tuberculosis Effects of Aminoguanidine on Latent Murine

1 downloads 91 Views 737KB Size Report
Einstein College of Medicine, 111 East 210th St., Bronx, NY 10467. E-mail address: ... Tower, Room E1240, Pittsburgh, PA 15261. E-mail address: ... Fresh lung tissue (three mice were analyzed for each time point) was flash- frozen in ..... Flynn, J. L., M. M. Goldstein, K. J. Triebold, J. Sypek, S. Wolf, and B. R. Bloom. 1995.
Effects of Aminoguanidine on Latent Murine Tuberculosis1 JoAnne L. Flynn,2* Charles A. Scanga,* Kathryn E. Tanaka,† and John Chan2‡§ A unique feature of Mycobacterium tuberculosis is its ability to establish latent infection in the human host, which can reactivate to cause disease years later. In the present study, the mechanisms involved in the control of latent tuberculous infection were examined using two murine experimental tuberculosis models. Analysis of the model involving infection of mice with a relatively low inoculum of the virulent Erdman strain of M. tuberculosis indicated that in vivo inhibition of reactive nitrogen intermediate (RNI) production by the nitric oxide synthase inhibitor aminoguanidine resulted in reactivation. This reactivation was evidenced by hepatosplenomegaly, a robust tissue granulomatous reaction, and increased bacillary load. IFN-g, TNF-a, and inducible nitric oxide synthase were all expressed throughout the latent phase of infection. Reactivation of latent tuberculous infection by aminoguanidine treatment was confirmed using a second murine tuberculosis model based on treatment with antimycobacterial drugs. Results obtained using this drug-based model also suggested the existence of an RNI-independent antimycobacterial mechanism(s) operative in the latent phase of infection. Together, these data suggest that both RNI-dependent and -independent mechanisms contribute to the prevention of tuberculous reactivation. The Journal of Immunology, 1998, 160: 1796 –1803.

T

uberculosis can manifest itself at different stages of infection. In primary infection, the clinical course ranges from benign self-limiting disease to progressive dissemination (1). It is thought that during acute primary tuberculosis, latent foci harboring dormant Mycobacterium tuberculosis are established in the host via hematogenous dissemination from the original site of infection (2). Recrudescence of these quiescent foci months to years later causes a disease entity known as chronic tuberculosis (2), which accounts for the majority of adult tuberculous infection in the United States (3). Despite the significance of reactivation in the pathogenesis of tuberculosis, the mechanisms involved in the development of latent infection and in subsequent recrudescence to active disease are not well understood. The basis for the reactivation theory is largely derived from clinical studies of tuberculosis (reviewed in Refs. 2 and 4). In chronic tuberculosis, the apical lung segments are by far the most common sites of active disease (2). Remarkably, apical scarring could be detected in a significant number of older persons long before the development of active pulmonary tuberculosis (4). These clinical data support the concept that chronic tuberculosis in adults, particularly the older population, is generally caused by reactivation of viable bacilli harbored in dormant foci in the apexes of the lung (2, 4). More importantly, clinical pathologic studies have shown that apical scars in humans without active tuberculosis contain cultivable tubercle bacilli, and such areas of pulmonary fibrosis can progress to active disease (2, 4).

*Departments of Molecular Genetics and Biochemistry, and Medicine, University of Pittsburgh School of Medicine, Pittsburgh, PA 15261; and Departments of †Pathology, ‡Medicine, and §Microbiology and Immunology, Albert Einstein College of Medicine, Bronx, NY 10467 Received for publication July 9, 1997. Accepted for publication October 10, 1997. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. 1 This work was supported in part by National Institutes of Health Grant AI36990 (to J.C. and J.L.F.) and the American Lung Association (to J.L.F.). 2 Address correspondence and reprint requests to Dr. John Chan, Departments of Medicine and Microbiology and Immunology, Montefiore Medical Center, Albert Einstein College of Medicine, 111 East 210th St., Bronx, NY 10467. E-mail address: [email protected]; or Dr. JoAnne L. Flynne, Department of Molecular Genetics and Biochemistry, University of Pittsburgh School of Medicine, Biomedical Science Tower, Room E1240, Pittsburgh, PA 15261. E-mail address: [email protected]

Copyright © 1998 by The American Association of Immunologists

It is well established that cell-mediated immunity is critical in host defense against M. tuberculosis (reviewed in Ref. 5). From clinical evidence, it is generally accepted that suppression of cellular immunity associated with corticosteroid therapy can result in reactivation of latent tuberculous infection (4, 6). The correlation between cell-mediated immunity and tuberculosis is, however, best illustrated by the remarkable susceptibility of individuals with AIDS to M. tuberculosis (7–9). Since reactivation plays a significant role in the pathogenesis of tuberculosis in HIV-infected persons, whose T cell-mediated immune responses are severely compromised, it is likely that attenuation of cell-mediated antimycobacterial mechanisms can lead to recrudescence of latent tuberculous infection. A potent cell-mediated antimycobacterial mechanism is effected via macrophage inducible nitric oxide synthase (iNOS)3 production of toxic reactive nitrogen intermediates (RNI). This L-arginine-dependent antimicrobial mechanism of macrophages has been established to be effective against M. tuberculosis both in vitro (10, 11) and in vivo (12–14). Thus, inhibition of iNOS function by chemical inhibitors such as NG-monomethyl L-arginine (NMMA) or aminoguanidine (AG) during acute M. tuberculosis infection led to fulminant and rapidly fatal disease progression associated with heavy bacterial burden in the lung, liver, and spleen (12). In addition, disruption of the function of TNF-a and IFN-g, the two cytokines critical for signaling RNI production by macrophages, compromised host defense against M. tuberculosis. Mice with disruptions in the genes for IFN-g or the 55-kDa TNF receptor were unable to produce RNI early in infection and quickly succumbed to M. tuberculosis infection (13, 14). Although the importance of these immune responses for control of acute tuberculous infection has been established, the contribution of macrophage antimicrobial RNI production to containing latent tuberculous infection has not been examined. Using two murine experimental tuberculosis models, the present study provides evidence suggesting that RNI play a significant role in the prevention

3 Abbreviations used in this paper: iNOS, inducible nitric oxide synthase; RNI, reactive nitrogen intermediates; NMMA, NG-monomethyl L-arginine; AG, aminoguanidine; PPD, purified protein derivative; AGE, advanced glycosylation end products; H&E, hematoxylin and eosin.

0022-1767/98/$02.00

The Journal of Immunology of tuberculousreactivation. In addition, the results suggest the existence of an RNI-independent antimycobacterial mechanism(s) that participates in the containment of latent tuberculosis.

Materials and Methods Animals Eight- to ten-week-old C57BL/6 female mice (The Jackson Laboratory, Bar Harbor, ME) were maintained in specific pathogen-free facilities. All experiments were performed in biosafety level 3 animal laboratories.

Mycobacteria and infection M. tuberculosis strain Erdman was prepared and maintained as previously described (10, 13, 14). Bacterial stocks, harvested from tissues of infected mice and expanded once (first passage), were stored at 280°C until use. Infection of mice was achieved i.v., via tail vein, at a dose of 5 3 103 to 1 3 104 viable CFU. For the low dose model, AG (2.5%, w/v) was given ad libitum in drinking water (12) beginning 6 mo postinfection. For the antimycobacterial drug-based model, mice were treated for 1 mo with isoniazid (0.1 g/l) and pyrazinamide (15 g/l) in drinking water ad libitum beginning 4 wk postinfection. AG treatment (2.5%, w/v, in drinking water) was initiated 1 wk after completion of antibiotic therapy. In both models, M. tuberculosis-infected control mice received plain drinking water. During the course of the study, moribund animals were killed to avoid suffering and were scored as succumbing to M. tuberculosis infection.

Chemicals and reagents All chemicals were purchased from Sigma Chemical Co. (St. Louis, MO). 7H9 medium and 7H10 agar substrates were obtained from Difco (Detroit, MI). Reagents for histopathologic studies and the RT-PCR were purchased from Vector Laboratories, Inc. (Burlingame, CA), and Life Technologies (Gaithersburg, MD), respectively.

Quantitation of viable bacilli in infected tissues Portions (25%) of the lung, liver, and spleen of killed mice were homogenized in PBS containing 0.05% Tween-80 (12–14). Dilutions of organ homogenates were plated on 7H10 agar plates, and CFU were counted following 21-day incubation at 37°C. There were three to five mice per group.

Histopathologic studies Tissue samples for histopathologic studies were prepared as previously described (12–14). Briefly, tissues were fixed in 10% buffered formalin before paraffin embedment. Hematoxylin and eosin and Ziehl-Neelsen acid fast staining of serial sections from paraffin blocks was examined to assess pathologic changes and bacillary load, respectively. The number and size of granulomas in five to seven random 103 fields was determined for each liver sample studied. The extent of granulomatous infiltration in infected liver tissues was assessed by the Cell Analysis Systems Micrometer version 1.0 program (Elmhurst, IL) according to the manufacturer’s instructions and was expressed as the granulomatous index, which indicates the ratio of area occupied by granulomas to the total area examined.

Immunohistochemical staining Formalin-fixed, paraffin-embedded tissues were used for immunohistochemical studies by the method of microwave Ag retrieval (15, 16). Preliminary studies have shown that this method yields the best results for staining by an affinity-purified rabbit polyclonal Ab against murine macrophage iNOS (provided by Dr. Charles Lowenstein, Johns Hopkins School of Medicine, Baltimore, MD) (14). Paraffin-embedded tissues were sectioned at 5 to 6 mm, deparaffinized, and subjected to microwave treatment in citrate buffer (pH 6.0) before immunostaining (15, 16). Anti-iNOS Abs were applied at a dilution of 1/100 for 4 h and detected using the avidin-biotin complex-based method according to the supplier’s instructions (Vector Laboratories); peroxidase-diaminobenzidine was used for development. The sections were counterstained with hematoxylin.

RT-PCR for IFN-g, TNF-a, and iNOS gene expression in tissues Expression of IFN-g, TNF-a, and iNOS in lung tissue of mice was examined by quantitative competitive RT-PCR as described previously (14, 17). Fresh lung tissue (three mice were analyzed for each time point) was flashfrozen in liquid nitrogen and stored at 280°C until use. RNA was prepared by homogenizing frozen tissue in Trizol according to the manufacturer’s protocol and treated with DNase to remove any contaminating DNA. RNA

1797 (5 mg) was reverse transcribed using SuperScript II (Life Technologies, Grand Island, NY). Quantitative competitive RT-PCR was performed as described previously (14, 17). Briefly, cDNA was standardized with primers for hypoxanthine-guanine phosphoribosyl transferase using a competitor plasmid that contains sequences for a number of different genes (18). Following standardization, equivalent amounts of cDNA were used in competitive PCR, using primers for each gene of interest (IFN-g, TNF-a, and iNOS) (18). PCR products were resolved on agarose gels and visualized with ethidium bromide.

Statistical analysis To ensure the normality assumption of the analysis, spleen, liver, and lung CFU were transformed using log base 10. These datasets were subjected to analysis by SAS program version 6.12 (SAS Institute, Cary, NC), using the general linear model.

Results and Discussion The murine latent tuberculosis models There are basically two murine models that have been described for the study of latent tuberculosis. The first involves a relatively low dose M. tuberculosis infection (the low dose model) that is solely controlled by the host’s immune response and remains quiescent for 15 to 18 mo, at which time reactivation occurs, presumably due to the immunocompromised state associated with aging (19). The second model uses a short course of antituberculous drugs (the drug model) to reduce the bacterial burden to a very low or undetectable level in the infected host (20 –24); the infection in these mice can reactivate upon cessation of antibiotics (20 –24) or upon treatment with immunosuppressive drugs such as glucocorticoid (22–24). Although neither is an exact replica of latent tuberculosis in man, each model possesses attributes that mimic human latency. In the low dose model, the host immune response is solely responsible for controlling the initial infection and maintaining the bacterial load at a stable level for many months, a situation similar to that in human latent tuberculosis. The weakness of this model is the relatively high bacillary burden in the spleen and lung of infected mice (Fig. 1). By contrast, the very low bacterial load in the organs of infected animals in the drug model is more akin to that in human latency; however, the effect of antibiotics on the initial host-bacterial interaction may be a confounding factor. Given that no true animal latency model exists, both the low dose and the drug-based models were used in this study to elucidate the mechanisms involved in the containment of latent tubercu losis. For the low dose model, C57BL/6 mice were infected i.v. with 5 3 103 to 1 3 104 of the virulent M. tuberculosis strain Erdman bacilli. Characterization of the kinetics of disease progression in this model (Fig. 1A) indicated that the liver is the most resistant organ to M. tuberculosis, and the lung is the most susceptible, as previously noted (25). Most importantly, there is an extended period, beginning 1 mo postinfection and extending for at least 10 mo, during which the infected host appears clinically healthy, and the bacillary burden is stably maintained. We reasoned that this is a suitable phase of infection in which to study latent tuberculosis. Indeed, in a related model, studies performed to examine the hostbacterium equilibrium 3 to 6 mo after infection with virulent M. tuberculosis provided evidence suggesting that during this period, bacilli were viable but not multiplying in the tissues (26). For the drug model, C57BL/6 mice were infected as described above. Four weeks following infection, mice received a 1-mo course of antimycobacterial therapy with isoniazid and pyrazinamide. This treatment resulted in the reduction of bacterial burden in all tissues examined to #200 bacilli/organ (see below). This low level of bacillary load was maintained for at least 6 mo after completion of antimycobacterial treatment.

1798

REACTIVATION OF LATENT TUBERCULOSIS The NOS inhibitor AG reactivates latent tuberculosis

FIGURE 1. Effect of AG on M. tuberculosis infection in the low dose murine model of latency. A, Kinetics of infection. C57BL/6 mice were infected i.v. with 5 3 103 to 1 3 104 CFU of the virulent Erdman strain of M. tuberculosis. At the indicated time intervals up to 10 mo postinfection, viable organisms in the lung (closed circles), spleen (open squares), and liver (open triangles) were determined by plating organ homogenates for CFU. Three mice per time point were used. Error bars indicate the SE. B, AG treatment of latently infected mice results in disease reactivation. M. tuberculosis-infected C57BL/6 mice were treated with AG (closed circles) beginning 6 mo postinfection (indicated as day 0 on graphs). AG was delivered in drinking water (2.5%, w/v) ad libitum. Control groups received plain drinking water (open squares). CFU in the lung (top panel), liver (middle panel), and spleen (bottom panel) were determined at various time points. Three mice per time point were used. Error bars indicate the SE. The experiment was performed twice with similar results. *, Indicates p 5 0.0001; **, p 5 0.0036; 1, p 5 0.0002; 11, p 5 0.0006.

The NOS inhibitor AG was used as a pharmacologic probe (12) to assess the role of toxic nitrogen oxides in localizing M. tuberculosis during the quiescent phase of infection. AG was used in this study because it is the most cost-effective of the various commonly used iNOS inhibitors. More importantly, by comparing the effects of AG on murine tuberculosis with those of the chemically distinct NMMA, we have previously established the validity of using either one of these NOS inhibitors in investigating the role of RNI in defense against M. tuberculosis infection in mice (12). One advantage of NMMA is that it allows correlation of in vivo RNI production to disease progression, because this NOS inhibitor has the capacity to abrogate virtually all RNI production in acute murine tuberculosis (12). By contrast, we and others (12, 27, 28) have demonstrated that reduction of nitrogen oxide production in vivo by AG can be incomplete, probably because of the relatively selective effect of AG against the inducible form of NOS. However, the cost of NMMA prohibited its use in these lengthy experiments. M. tuberculosis-infected mice treated with AG at 6 mo postinoculation developed reactivated tuberculosis in the low dose model. This AG-induced reactivation was associated with progressively fatal disease (mean survival time of 62 6 2 days post-AG treatment) as well as increased bacillary burden compared with those in control mice (Fig. 1B), all of which survived the experiment (12 mo). Interestingly, as is well described for acute murine tuberculosis, the kinetics of bacillary proliferation in the lung, liver, and spleen of the AG-reactivated mice followed distinct patterns (Fig. 1B). The effect of AG-induced reactivation on bacterial burden is most prominent in the lung. The kinetics of pulmonic mycobacterial multiplication were progressive, with CFU reaching as high as 109 at later time points (1000-fold more than that in the lung of control mice). In contrast, the number of hepatic CFU in AG-treated mice did not increase until 8 wk post-AG treatment, suggesting that the liver is relatively resistant to reactivated M. tuberculosis infection. The kinetics of reactivation observed in the spleen followed yet another pattern. Unlike the rapidly progressive expansion of CFU seen in the lung, splenic bacterial multiplication followed a much slower course, even though the bacillary loads of these two organs at the time of initiation of AG treatment were comparable. AG treatment also evoked pathologic responses closely resembling those of an acute infection. Significantly, as early as 2 wk after initiation of AG treatment, mice in the latent phase of infection developed hepatosplenomegaly, a characteristic finding in acute tuberculosis. The most prominent component of these host responses was the granulomatous reaction, the hallmark of tuberculosis. Upon AG treatment, the granulomas enlarged and became more structured and cellular (Fig. 2, A–D). In tissues obtained from infected mice treated with AG, the area occupied by granulomas far exceeded that in untreated controls (Table I). This AGinduced granulomatous reaction was most apparent in the liver and least apparent in the lung. The discrepancy is probably related to the numbers of bacilli in the tissues studied during the latent phase of infection. To contain the infection foci in the lung, which carried a higher bacillary load compared with that of the liver (Fig. 1A), a relatively vigorous tissue response was maintained, even at the quiescent state of infection. As a result, augmentation of this degree of tissue reaction secondary to AG-induced reactivation may not be readily appreciated. Nevertheless, the effect of AG on the granulomatous reaction in the pulmonic tissues of mice with quiescent disease became obvious at later time points: by 10 wk post-AG treatment, lung granulomas of AG-reactivated mice had progressed remarkably(Fig. 2F), and on examining multiple tissue sections, there was evidence of necrosis. In contrast, the vigorous

The Journal of Immunology

1799

FIGURE 2. Histopathologic studies of tissues from latently infected, AG-treated mice in the low dose model. Mice were infected for 6 mo before the initiation of AG treatment (see Fig. 1). A through D, H&E-stained liver sections from control infected (A and C) and AG-treated infected (B and D) mice. Two weeks of AG treatment induced a marked granulomatous response in the livers of latently infected mice (A and B). By 6 wk after initiation of AG treatment (C and D), this granulomatous response began to subside. E and F, H&E-stained lung sections from control infected (E) and AG-treated infected (F) mice at 10.5 wk post-treatment. Lung tissue in AG-treated mice (F) shows an extensive granulomatous response. Magnifications, 3200 (insets, 3800).

AG-induced granulomatous reaction in the liver, which suggests hepatic recrudescence of tuberculosis, began to subside at 6 wk post-treatment (Fig. 2, C and D). This latter observation suggests that the liver is uniquely resistant to M. tuberculosis during the reactivated phase of infection. Together, these results provide evidence that in the low dose model, RNI may play a role in the containment of latent tuberculosis.

To explore the possibility that AG-induced reactivation of latent tuberculosis was due to a direct adverse effect of AG on T cell response, an important component of host defense against M. tuberculosis, we performed in vitro functional assays on lymphocytes explanted from infected, AG-treated mice. Con A- and PPDinduced proliferation of splenocytes from these mice was comparable to that of untreated controls (data not shown). In fact,

Table I. Granulomatous reaction is increased in liver following inhibition of iNOS function in latently infected mice a Inner Area (mm)

No. Granulomas

Granulomatous Ratiob

Weeks Post-AG Treatment

2AG

1AG

2AG

1AG

2AG

1AG

2 6 10.5

1.6 6 1.3 1.0 6 0.9 0.8 6 0.7

6.9 6 2.2* 1.9 6 1.1† 2.0 6 1.2†

4956 6 4186 2628 6 678.6 1976 6 2471

45110 6 26240* 3642 6 835.5‡ 5201 6 3500†

1.0 6 0.8 0.4 6 0.5 0.4 6 0.5

8.8 6 5.1* 0.9 6 0.6‡ 1.0 6 0.7†

a

Each value is the mean value per 103 field; five to seven fields per section were examined. There were three mice per group at each time point. Ratio of granulomatous area to nongranulomatous area of tissue. *p # 0.0001; † p # 0.006; ‡ p # 0.05. b

1800

REACTIVATION OF LATENT TUBERCULOSIS

FIGURE 3. iNOS protein expression in latent and reactivated tuberculous mice. Liver (A–D) and lung (E and F) sections were stained for iNOS using anti-iNOS Ab, as described. At 2 wk post-AG treatment, iNOS staining, which is primarily in the granulomas, was much more pronounced in AG-treated mice (B) than that in infected controls (A). By 6 wk post-AG treatment, the iNOS expression in AG-treated mice (D) had regressed to levels comparable to those in controls (C). Inducible NOS expression was noted at all time points in lung sections from latently infected mice (E and data not shown). An AG-induced increase in pulmonic iNOS expression (F), compared with that in controls (E), was apparent only at the later time points (10.5 wk post-AG treatment). Tissue sections from uninfected mice did not stain positively with the iNOS-specific Ab used in this study (data not shown). Ab staining is brown. Magnification, 3200.

the PPD-induced proliferative response was higher in the AGtreated mice. In addition, splenocytes from these two groups of mice produced equivalent amounts of IFN-g, as measured by ELISA, following in vitro stimulation with Con A and PPD (data not shown). These results suggest that the AG-induced reactivation of latent tuberculosis is not due to suppression of the T cell response. Rather, AG induces recrudescence of quiescent disease via direct suppression of RNI production by the macrophage L-arginine-dependent antimicrobial mechanism. Expression of iNOS in latent and reactivated tuberculosis To further characterize the role of RNI in the containment of M. tuberculosis in the low dose model, tissue expression of iNOS was assessed by immunohistochemical studies using an affinity-purified polyclonal Ab against murine macrophage iNOS (14). The results of these studies revealed that iNOS is expressed in all latently infected tissues examined, suggesting that RNI are necessary for the control of latent M. tuberculosis infection (Fig. 3).

An increase in tissue iNOS protein levels was observed upon treatment with AG (Fig. 3). This AG-induced, reactivation-associated increase in iNOS expression was most apparent in the liver (Fig. 3, A–D). While enhancement of iNOS expression in the liver was remarkable after 2 wk of AG treatment, that in the lung was apparent only at later times (Fig. 3). This discrepancy may be due to the higher baseline level of iNOS expression in the lung compared with that in the liver during the latent phase of infection, thus rendering a further increase in pulmonic expression not readily apparent. Interestingly, at 6 wk after initiation of AG treatment, when the regression of hepatic granulomatous reaction began (Fig. 2, C and D), iNOS expression in the liver was also noted to start decreasing (Fig. 3). These results further suggest that the liver is unique in its ability to adequately control AG-reactivated tuberculosis in the low dose model. Since AG is effective in attenuating RNI production in vivo in various animal models of infectious diseases, including tuberculosis (12, 28), the relative resistance of the liver to M. tuberculosis in the reactivation

The Journal of Immunology

1801

FIGURE 4. The iNOS, IFN-g, and TNF-a mRNA expression in latent tuberculous mice in the low dose model. At 6.5 and 7.5 mo postinfection, lung RNA was obtained from C57BL/6 mice infected or uninfected (UN) with 104 CFU of the virulent Erdman strain of M. tuberculosis i.v. Competitive RT-PCR on cDNA from three mice from each group was performed, and the results are shown here. Samples were standardized for hypoxanthine guanine phosphoribosyl transferase (HPRT) mRNA, and then used in competitive reactions with primers specific for iNOS, IFN-g, and TNF-a (36). The competitor plasmid concentration was the same in all reactions. In each panel, the upper band represents plasmid competitor DNA, and the lower band represents tissue cDNA. Messages of iNOS, IFN-g, and TNF-a were detectable from samples obtained from infected mice at all time points tested, but not from uninfected animals.

phase of the infection may be due to macrophage iNOS-independent antimycobacterial mechanisms. Finally, the expression of iNOS during the quiescent phase of infection in the low dose model of latent tuberculosis is further reinforced by the detection of iNOS mRNA in latently infected tissue for up to at least 8 mo postinfection by competitive RT-PCR analysis (Fig. 4). Expression of cytokines in latent tuberculosis IFN-g and TNF-a are the key factors that activate the macrophage RNI-generating pathway (29) and are essential in the control of acute tuberculous infection (13, 14, 30). Therefore, we evaluated the expression of these cytokines in the quiescent phase of tuberculous infection. Examining the lung tissue of infected mice, results obtained by RT-PCR indicated that both IFN-g and TNF-a are expressed throughout the quiescent phase of tuberculous infection (Fig. 4), suggesting that these cytokines participate in the control of latent tuberculosis at least in part by maintaining the RNI-generating pathway in an activated state. Indeed, it was shown recently that neutralization of TNF-a by administration of an adenovirus vector expressing the extracellular domain of the human p55 TNF receptor exacerbated acute and chronic M. tuberculosis infection in mice (31). In summary, the results demonstrating the expression of iNOS, IFN-g, and TNF-a in latently infected tissues together with the ability of AG to trigger clinical reactivation, as manifested by rapid development of hepatosplenomegaly, a marked granulomatous response, and increased bacillary burden, strongly suggest that continuous production of RNI at latent foci of infection plays an important role in preventing tuberculous reactivation. AG treatment also reactivates infection in the antimycobacterial drug model of latent tuberculosis. To more stringently test the role of RNI in preventing tuberculous reactivation, we examined the

FIGURE 5. AG treatment in the antimycobacterial drug model of latent tuberculosis. M. tuberculosis-infected C57/BL6 mice were treated with isoniazid and pyrazinamide for 4 wk. One week after completion of chemotherapy, AG treatment was started (indicated as day 1). Mice were given either plain drinking water (open squares) or water containing 2.5% (v/w) AG (close circles). The bacterial load from organs was determined at various time points post-AG treatment. The experiment was performed twice; data from one representative experiment are shown here. Three to five mice per time point were killed. Error bars show the SE. *, indicates p 5 0.0001; **, p 5 0.0095; 1, p 5 0.0051; 11, p 5 0.0086. The effect of time on CFU of controls (open squares) is not significant (lung, p 5 0.6694; liver, p 5 0.0615; spleen, p 5 0.0712).

effect of AG on mice latently infected with M. tuberculosis using the drug model. As in the case of the low dose model, AG treatment resulted in increased bacterial growth in the lung, liver, and spleen (Fig. 5). Again, the largest increase in bacillary burden was seen in the lung, where AG treated mice had a 1000-fold more viable CFU than controls by 85 days post-treatment (Fig. 5). The bacterial burden in control mice was maintained at 102 to 103 CFU/organ during the study period (24 wk). Interestingly, after an initial rise in pulmonary CFU to ;105/organ, AG-treated mice in the drug model were able to stably maintain this bacillary burden for .80 days (Fig. 5). By contrast, this ability of the lung to apparently stabilize bacterial multiplication was not seen in the low dose model. In the low dose model, AG-treated mice could not control disease progression, resulting in an increase in pulmonary CFU from ;6 3 105 to 109 within 80 days after AG treatment was

1802 initiated (Fig. 1). These data suggest the existence of an iNOSindependent antimycobacterial mechanism(s) that may contribute to the control of latent murine tuberculosis. The recent in vitro demonstration that various strains of M. tuberculosis may vary widely in their susceptibility to the toxic effects of RNI provides indirect evidence that these latter mechanisms exist (32, 33). Given the wide array of redundant host defense strategies, the existence of an RNI-independent antimycobacterial mechanism(s) operative in the control of latent tuberculosis should not be difficult to envision. Although not yet defined, these iNOS-independent mechanisms may be related to IFN-g and TNF-a, given the fact that these two cytokines, which are known to play critical roles in host defense against M. tuberculosis, are expressed throughout the quiescent phase of tuberculous infection. Commentary The macrophage L-arginine-dependent cytotoxic pathway mediates a major antimycobacterial mechanism in acute murine tuberculosis via the generation of toxic RNI (10 –14). Importantly, iNOS has been shown to be highly expressed in bronchial alveolar macrophages obtained from pulmonary lavage fluids of patients infected with M. tuberculosis (34), suggesting a role for RNI in host defense against the tubercle bacillus in humans. Directly relevant to this study, the NOS inhibitor N6-(1-iminoethyl)-L-lysine has been shown to induce tubercular progression when administered to mice 1 mo after infection with 105 CFU of the virulent Erdman strain of M. tuberculosis (35). Although not confirmed by the drug model, these latter results support the evidence provided by the present study suggesting that RNI contribute to the control of latent tuberculosis. In addition, macrophage iNOS has been demonstrated to participate in controlling latent murine leishmaniasis (36). Using the two models of latent murine tuberculosis currently available, results from the present study suggest that both RNI-dependent and -independent antimycobacterial mechanisms contribute to the prevention of disease reactivation. Detailed analyses of the low dose model indicated that the three critical factors whose concerted actions culminate in the generation of NO and related nitrogen oxides, IFN-g, TNF-a, and iNOS, are all expressed during the latent phase of tuberculous infection. Attenuation of the production of mycobacteriocidal RNI leads to reactivation of latent infection, evidenced by increases in bacillary burden and mortality. This reactivation phase is associated with pathologic findings reminiscent of acute disease, hepatosplenomegaly and a vigorous granulomatous reaction, with concomitant enhanced expression of the RNI-generating enzyme iNOS. These results indicate that attenuation of RNI production in latent tuberculosis results in disease recrudescence. The role of RNI in controlling latent tuberculous infection is confirmed using the drugbased model. Interestingly, data derived from the drug model suggest that an iNOS-independent antimycobacterial mechanism(s) also contributes to the prevention of reactivation (Figs. 1 and 5). Thus, the two models of murine latent tuberculosis employed in this study may afford a useful system for comparative analysis of antimycobacterial functions operative during latent tuberculous infection. Because of the lack of true latent animal tuberculosis, these two models, which, in the strictest sense, may only represent a chronic persistent form of tuberculous infection, are an approximation of human latency at best. Therefore, the significance of the observations of this study to human latent tuberculosis remains to be rigorously tested. Although we have demonstrated that AG is a suitable iNOS inhibitor for examining the role of RNI in murine tuberculosis (12), its known biologic targets other than iNOS should not be dismissed lightly. These include the potential of AG to interfere

REACTIVATION OF LATENT TUBERCULOSIS with the metabolism of immunoregulatory polyamines (37) and to chemically modulate certain ligands for the macrophage scavenger receptor (38). However, the significance of these attributes of AG in vivo is unclear. Indeed, AG probably does not significantly affect the intracellular balance of polyamines in vivo, since there is evidence that a compensatory increase in excretion of unmodified polyamines occurs during AG administration, thereby preventing intracellular accumulation (39). The relevance of the ability of AG to modify the ligands for macrophage scavenger receptors is no clearer. In hyperglycemic states, such as in a diabetes model, AG acts as an inhibitor of the formation of advanced glycosylation end products (AGEs) (40). The AGEs are modified biologic proteins, including certain ligands for macrophage scavenger receptors, whose formation is accelerated in the presence of high glucose levels. The ability of AG to inhibit the formation of AGEs in a nonhyperglycemic system, such as our murine latent tuberculosis models, is entirely unknown. Regardless of the in vivo significance of these side effects of AG and other NOS inhibitors, the caveats of using NOS inhibitors as a means to attenuate NO production in any biologic system are noteworthy. Latent tuberculous infection plays a significant role in the pathogenesis of M. tuberculosis (2, 4), a pathogen that causes 3 million deaths annually worldwide (41, 42). Given that one-third of the world’s population is currently infected with M. tuberculosis, hosts chronically harboring tubercle bacilli in the latent phase constitute a significant reservoir of tuberculosis. These individuals can serve as a source of disease dissemination when reactivation of latent infection occurs. Understanding the mechanisms involved in the development of latent and reactivation tuberculosis is, therefore, of paramount importance to the prevention, treatment, and control of tuberculosis.

Acknowledgments We are grateful to Dr. Charles Lowenstein for providing iNOS Ab, and to Dr. Simon Watkins for imaging and photography assistance. We thank Drs. Barry Bloom and John McKinney for critical reading of the manuscript, Amy Myers for technical assistance, and Ming Tsang for immunohistochemical staining.

References 1. Stead, W. W., G. R. Kerby, D. P. Schleuter, and C. W. Jordahl. 1968. The clinical spectrum of primary tuberculosis in adults: confusion with reinfection in the pathogenesis of chronic tuberculosis. Ann. Intern. Med. 68:731. 2. Stead, W. W. 1967. Pathogenesis of a first episode of chronic pulmonary tuberculosis in man: recrudescence of residuals of the primary infection or exogenous reinfection? Am. Rev. Respir. Dis. 95:729. 3. Anonymous. 1990. The use of preventive therapy for tuberculous infection in the United States: recommendations of the Advisory Committee for Elimination of Tuberculosis. Morb. Mortal. Wkly. Rep. 39:9. 4. Stead, W. W. 1965. The pathogenesis of pulmonary tuberculosis among older persons. Am. Rev. Respir. Dis. 91:811. 5. Chan, J., and S. H. E. Kaufmann. 1994. Immune mechanisms of protection. In Tuberculosis: Pathogenesis, Protection and Control. B. R. Bloom, ed. American Society for Microbiology, Washington, DC, pp. 389 – 415. 6. Haas, D. W., and R. M. des Perez. 1995. Mycobacterium tuberculosis. In Principles and Practice of Infectious Diseases. G. L. Mandell, J. E. Bennett, and R. Dolin, eds. Churchill Livingstone, New York, pp. 2213–2243. 7. Snider, D. E. J., and W. L. Roper. 1992. The new tuberculosis. N. Engl. J. Med. 326:703. 8. Barnes, P. F., A. B. Bloch, P. T. Davidson, and D. E. Snider, Jr. 1991. Tuberculosis in patients with human immunodeficiency virus infection. N. Engl. J. Med. 234:1644. 9. Selwyn, P. A., D. Hartel, V. A. Lewis, E. E. Schoenbaum, S. H. Vermund, R. S. Klein, A. T. Walker, and G. H. Freidland. 1989. A prospective study of the risk of tuberculosis among intravenous drug users with human immunodeficiency virus infection. N. Engl. J. Med. 320:545. 10. Chan, J., Y. Xing, R. Magliozzo, and B. R. Bloom. 1992. Killing of virulent Mycobacterium tuberculosis by reactive nitrogen intermediates produced by activated murine macrophages. J. Exp. Med. 175:1111. 11. Denis, M. 1991. Interferon-gamma-treated murine macrophages inhibit growth of tubercle bacilli via the generation of reactive nitrogen intermediates. Cell. Immunol. 132:150.

The Journal of Immunology 12. Chan, J., K. Tanaka, D. Carroll, J. L. Flynn, and B. R. Bloom. 1995. Effect of nitric oxide synthase inhibitors on murine infection with Mycobacterium tuberculosis. Infect. Immun. 63:736. 13. Flynn, J. L., J. Chan, K. J. Triebold, D. K. Dalton, T. A. Stewart, and B. R. Bloom. 1993. An essential role for Interferon-g in resistance to Mycobacterium tuberculosis infection. J. Exp. Med. 178:2249. 14. Flynn, J. L., M. M. Goldstein, J. Chan, K. J. Triebold, K. Pfeffer, C. J. Lowenstein, R. Schreiber, T. W. Mak, and B. R. Bloom. 1995. Tumor necrosis factor-a is required in the protective immune response against M. tuberculosis in mice. Immunity 2:561. 15. Swanson, P. E. 1994. Microwave antigen retrieval in citrate buffer. Lab. Med. 25:520. 16. Cattoretti, G., M. H. G. Becker, and G. Key. 1992. Monoclonal antibodies against recombinant parts of the Ki-678 antigen (MIB1 and MIB3) detect proliferating cells in microwave-processed formalin-fixed tissue sections. J. Pathol. 168:357. 17. Flynn, J. L., M. M. Goldstein, K. J. Triebold, J. Sypek, S. Wolf, and B. R. Bloom. 1995. IL-12 increases resistance of BALB/c mice to Mycobacterium tuberculosis infection. J. Immunol. 155:2515. 18. Reiner, S. L., S. Zheng, D. B. Corry, and R. M. Locksley. 1993. Constructing polycompetitor cDNAs for quantitative PCR. J. Immunol. Methods 165:37. 19. Orme, I. M. 1988. A mouse model of the recrudescence of latent tuberculosis in the elderly. Am. Rev. Respir. Dis. 137:716. 20. McCune, R. M., and R. Tompsett. 1957. Fate of Mycobacterium tuberculosis in mouse tissues as determined by the microbial enumeration technique. I. The persistence of drug-susceptible tubercle bacilli in the tissues despite prolonged antimicrobial therapy. J. Exp. Med. 104:737. 21. McCune, R. M., R. Tompsett, and W. McDermott. 1957. The fate of Mycobacterium tuberculosis in mouse tissues as determined by the microbial enumeration technique. II. The conversion of tuberculous infection to the latent state by the administration of pyrazinamide and a companion drug. J. Exp. Med. 104:763. 22. McCune, R. M., F. M. Feldmann, H. P. Lambert, and W. McDermott. 1966. Microbial persistence. I. The capacity of tubercle bacilli to survive sterilization in mouse tissues. J. Exp. Med. 123:445. 23. McCune, R. M., F. M. Feldmann, and W. McDermott. 1966. Microbial persistence. II. Characteristics of the sterile state of tubercle bacilli. J. Exp. Med. 123: 469. 24. Grosset, J. 1978. The sterilizing value of rifampicin and pyrazinamide in experimental short course chemotherapy. Tubercle 59:287. 25. Orme, I. M., and F. M. Collins. 1994. Mouse model of tuberculosis. In Tuberculosis: Pathogenesis, Protection, and Control. B. R. Bloom, ed. American Society for Microbiology, Washington, DC, pp. 113–134. 26. Rees, R. J. W., and D.’A. Hart. 1961. Analysis of the host-parasite equilibrium in chronic murine tuberculosis by total and viable bacillary counts. Br. J. Exp. Pathol. 42:83. 27. Tilton, R. C., K. Chang, J. A. Corbett, T. P. Misko, M. G. Currie, N. S. Bora, H. J. Kaplan, and J. R. Williamson. 1994. Endotoxin-induced uveitis in the rat is

1803

28.

29. 30.

31.

32.

33.

34.

35.

36.

37. 38.

39. 40.

41. 42.

attenuated by inhibition of nitric oxide production. Invest. Ophthal. Vis. Sci. 35:3278. Beckerman, K. P., H. W. Rogers, J. A. Corbett, R. D. Schreiber, M. L. McDaniel, and E. R. Unanue. 1993. Release of nitric oxide during the T cell independent pathway of macrophage activation. J. Immunol. 150:888. Nathan, C. 1992. Nitric oxide as a secretory product of mammalian cells. FASEB J. 6:3051. Cooper, A. M., D. K. Dalton, T. A. Stewart, J. P. Griffen, D. G. Russell, and I. M. Orme. 1993. Disseminated tuberculosis in IFN-g gene-disrupted mice. J. Exp. Med. 178:2243. Adams, L. B., C. M. Mason, J. K. Kolls, D. Scollard, J. L. Krahenbuhl, and S. Nelson. 1995. Exacerbation of acute and chronic murine tuberculosis by administration of a tumor necrosis factor receptor-expressing adenovirus. J. Infect. Dis. 171:400. O’Brien, L., J. Carmichael, D. B. Lowrie, and R. W. Andres. 1994. Strains of Mycobacterium tuberculosis differ in susceptibility to reactive nitrogen intermediates in vitro. Infect. Immun. 62:5187. Rhoades, E. R., and I. M. Orme. 1997. Susceptibility of panel of virulent strains of Mycobacterium tuberculosis to reactive nitrogen intermediates. Infect. Immun. 65:1189. Nicholson, S., M. Bonecini-Almeida, J. R. L. Silva, C. Nathan, Q.-w. Xie, R. Mumford, J. R. Weidner, J. Calaycay, J. Geng, N. Boechat, C. Linhares, W. Rom, and J. L. Ho. 1996. Inducible nitric oxide synthase in pulmonary alveolar macrophages from patients with tuberculosis. J. Exp. Med. 184:2293. MacMicking, J. D., R. J. North, R. LaCourse, J. S. Mudgett, S. K. Shah, and C. F. Nathan. 1997. Identification of nitric oxide synthase as a protective locus against tuberculosis. Proc. Natl. Acad. Sci. USA 94:5243. Stenger, S., N. Donhaur, H. Thuring, M. Rollinghoff, and C. Bogdan. 1996. Reactivation of latent leishmaniasis by inhibition of inducible nitric oxide synthase. J. Exp. Med. 183:1501. Seiler, N., B. Knodgen, and K. Haegele. 1882. N-(3-aminopropyl) pyrrolidin-2one, a product of spermidine catabolism in vivo. Biochem. J. 208:189. Kuzuya, M., S. Satake, H. Miura, T. Hayashi, and A. Iguchi. 1996. Inhibition of endothelial cell differentiation on a glycosylated reconstituted basement membrane complex. Exp. Cell Res. 226:336. Seiler, N., F. N. Bolkenius, and B. Knodgen. 1985. The influence of catabolic reactions on polyamine excretion. Biochem. J. 225:219. Bucala, R., K. J. Tracey, and A. Cerami. 1991. Advanced glycosylation products quench nitric oxide and mediate defective endothelium-dependent vasodilatation in experimental diabetes. J. Clin. Invest. 87:432. Bloom, B. R., and C. Murray. 1992. Tuberculosis: commentary on a reemergent killer. Science 257:1055. Dolin, P. J., M. C. Raviglione, and A. Kochi. 1994. Global tuberculosis incidence and mortality during 1990 –2000. Bull. W. H. O. 72:213.