Iron-cofactored Superoxide Dismutase Inhibits Host ... - ATS Journals

Iron-cofactored Superoxide Dismutase Inhibits Host Responses to Mycobacterium tuberculosis KATHRYN M. EDWARDS, MICHAEL H. CYNAMON, RAMA K. R. VOLADRI, CYNTHIA C. HAGER, MICHELLE S. DESTEFANO, KYI T. THAM, DAVID L. LAKEY, MARKIAN R. BOCHAN, and DOUGLAS S. KERNODLE Departments of Pediatrics, Medicine, and Pathology, Vanderbilt University School of Medicine; Veterans Affairs Medical Center, Nashville, Tennessee; SUNY Health Science Center and Veterans Affairs Medical Center, Syracuse, New York

Superoxide dismutase (SOD) is a ubiquitous metalloenzyme in aerobic organisms that catalyzes the conversion of superoxide anion to hydrogen peroxide. Mycobacterium tuberculosis is unusual in that it secretes large quantities of iron-cofactored SOD. To determine the role of SOD in pathogenesis, we constructed mutants of M. tuberculosis H37Rv with reduced SOD production. Compared with controls, SOD-diminished isolates were more susceptible to killing by hydrogen peroxide. The isolates were markedly attenuated, exhibiting nearly 100,000-fold fewer bacilli than virulent control strains in the lungs and spleens of C57BL/6 mice 4 wk after intravenous inoculation. In the lung, SOD-attenuated M. tuberculosis induced robust interstitial mononuclear cell infiltration within 24 h and many cells were apoptotic by TUNEL staining, whereas virulent H37Rv exhibited minimal early inflammatory response and only rare interstitial mononuclear cell apoptosis. During prolonged infections, C57BL/6 mice tolerated SOD-attenuated M. tuberculosis better than BCG, exhibiting 68% greater weight gain, quicker eradication of bacilli from the spleen, and less alveolar lung infiltration. These results establish the importance of SOD in the pathogenesis of tuberculosis. Its effect appears to be mediated in part by inhibiting innate host immune responses, including early mononuclear cell infiltration of infected tissues and apoptosis. Keywords: apoptosis; superoxide dismutase; tuberculosis

be achieved only if a second episomal copy of the allele is present (14). Mycobacterium tuberculosis produces two SODs, an iron-cofactored enzyme (SOD) encoded by sodA, and a copper-zinc enzyme encoded by sodC (15–18). Dussurget and coworkers reported that inactivation of sodC does not reduce M. tuberculosis virulence in guinea pigs (18). However, their attempts to inactivate sodA were unsuccessful, leading them to suggest that sodA might be essential for viability. Unlike nonpathogenic mycobacteria and most other bacterial species, pathogenic mycobacteria produce large amounts of iron-cofactored SOD, and much of it is exported extracellularly (15, 17). The production and export of 93- and 350-fold more SOD, respectively, by M. tuberculosis than the nonpathogenic mycobacterium M. smegmatis (17) poignantly exemplifies the magnitude of this difference. As M. tuberculosis is predominantly an intracellular pathogen during the early stages of infection, we hypothesized that the high-level production and export of SOD by tubercle bacilli might have a role in the pathogenesis of tuberculosis, and used antisense (AS) RNA expression to construct SOD-attenuated mutants to evaluate this question.

METHODS Construction of SOD-attenuated Mutants of H37Rv

Tuberculosis remains a major cause of human illness despite the availability of effective treatment regimens and the use of bacillus Calmette-Guérin (BCG) vaccine (1, 2). Furthermore, a reservoir comprising approximately one-third of the world’s population is currently infected. Advances in mycobacterial genetics make it possible to determine the roles of specific microbial factors in the pathogenesis of tuberculosis and to construct new vaccine candidates (3–10). One strategy for attenuating intracellular pathogens involves making them susceptible to host immune defense mechanisms by eliminating or reducing microbial factors that mediate intracellular survival (11). Thus, microbial factors involved in detoxifying macrophage-produced reactive oxygen and nitrogen intermediates are logical targets. Superoxide dismutase (SOD) is a ubiquitous metalloenzyme in aerobic organisms, including pathogenic and nonpathogenic species of mycobacteria. SOD detoxifies superoxide anion and contributes to the virulence of intracellular pathogens including Salmonella typhimurium and Yersinia enterocolitica (12, 13). In Legionella pneumophila, iron-cofactored SOD is essential for viability, and allelic inactivation of the chromosomal gene can

( Received in original form June 20, 2001; accepted in final form September 10, 2001) Supported by NIH RO1 AI37871 and a Merit Review Award from the Medical Research Service of the Department of Veterans Affairs. Correspondence and requests for reprints should be addressed to Douglas S. Kernodle, MD, Medical Service (111E), Nashville VA Medical Center, 1310 24th Avenue South, Nashville, TN 37212-2637. E-mail: [email protected] Am J Respir Crit Care Med Vol 164. pp 2213–2219, 2001 DOI: 10.1164/rccm2106093 Internet address: www.atsjournals.org

The pAL5000 origin of replication as a 2.6-kb DNA fragment from pBAK14 (15), a kanamycin resistance gene on a 1.2-kb fragment from pY6002 (19), and the promoter of the 65-kDa heat shock protein (PrHSP) polymerase chain reaction (PCR) amplified from M. tuberculosis chromosomal DNA were ligated into pBluescript SK (Stratagene, La Jolla, CA) to make pHV203 (Figure 1A). A 151-bp fragment of sodA corresponding to amino acids 139–189 of SOD was PCR amplified from chromosomal DNA and ligated into the multicloning site to make pHV203-AS-SOD. A second antisense-sodA plasmid, pLUC10AS-SOD, was constructed by replacing the proximal portion of the luciferase gene of pLUC10 (20) with the sodA fragment. In both vectors the sodA fragment was behind Pr-HSP in reverse orientation relative to its polarity on the chromosome. Plasmids were electroporated into M. smegmatis mc2155 (20) and M. tuberculosis H37Rv (ATCC 25618).

In Vitro Analyses Mycobacterial strains were cultivated in Middlebrook 7H9 liquid medium or 7H10 agar (Difco, Detroit, MI) supplemented with 0.2% glycerol and 10% OADC (oleic acid–albumin–dextrose) enrichment (Difco or Becton Dickinson [Franklin Lakes, NJ]) unless otherwise indicated, adding kanamycin at 25–50 g/ml for strains containing selectable markers. Growth rates were determined by monitoring changes in A600 over time. To perform Northern hybridizations, RNA from M. smegmatis transformants was electrophoresed in 1.5% agarose, transferred to nylon membranes, and hybridized. For Western hybridization, lysed bacterial cells were adjusted to a standard protein concentration, electrophoresed on a 12% polyacrylamide gel, and transferred to nitrocellulose membranes, which were hybridized with E-293 antisera against H37Rv whole cell lysate. To identify the location of SOD, membranes were stripped and hybridized with rabbit polyclonal antisera against recombinant M. tuberculosis SOD (21). Densitometric scanning of the SOD bands was used to quantify SOD production. Hydrogen peroxide killing assays were performed in triplicate in Middlebrook 7H9 broth containing albumin–dextrose and 0.2% glycerol but lacking oleic acid and catalase.

2214

AMERICAN JOURNAL OF RESPIRATORY AND CRITICAL CARE MEDICINE

VOL 164

2001

Figure 1. In vitro characteristics of H37Rv expressing an antisense sodA fragment. Map of the antisense expression vector pHV203 (A). It contains origins of replication for E. coli and mycobacteria, genes for chloramphenicol and kanamycin resistance, and a multicloning site (MCS) behind a strong promoter (Pr-HSP). The AS-sodA DNA fragment was cloned into the MCS. Northern hybridization of antisense sodA RNA transcript (B). Lane 1, M. smegmatis mc2155; lane 2, mc2155 containing a control plasmid; lane 3, mc2155 containing pLUC10-AS-SOD. Because of strong homology between sodA of M. tuberculosis and sodA of M. smegmatis, a probe for the luciferase RNA transcript was used to detect expression of antisense RNA transcribed from pLUC10-AS-SOD. The antisense sodA fragment is between the promoter and the probed transcript. Western hybridization of SOD (C). Lane 1, H37Rv; lane 2, H37Rv(pLUC10); lane 3, H37Rv(pLUC10-ASSOD); lane 4, recombinant M. tuberculosis SOD from E. coli, which is slightly larger because of a histidine tag fusion. An arrow identifies the band corresponding to SOD. Susceptibility to hydrogen peroxide (D). The log CFU of surviving H37Rv(pLUC10) (open columns) and H37Rv(pLUC10-ASSOD) (solid columns) bacilli are displayed as means SEM of triplicate determinations.

Murine Inoculation and Clearance Studies Female C57BL/6 mice (H-2b) aged 5–6 wk (Jackson Laboratories, Bar Harbor, ME) were maintained in a pathogen-free biosafety level-3 facility, with water and Prolab RMH 3000 rodent chow (Purina, St. Louis, MO) ad libitum. Animal procedures were approved by the Syracuse Veterans Affairs Medical Center (VAMC) Subcommittee on Animal Studies. Two 28-d experiments with six arms involving five strains were performed, with H37Rv(pLUC10-AS-SOD) used in both experiments. Each arm included 24 mice, divided into 4 groups of 6. An additional experiment 17 mo in duration, compared the Tice substrain of BCG (Organon Teknika, Durham, NC) and H37Rv(pLUC10-AS-SOD). Inocula were administered to mice by tail vein. Killing was achieved by CO2 inhalation. Spleens and right lungs were removed aseptically and disrupted in a sealed grinding assembly attached to a homogenizer, and bacterial counts were titered on nonselective agar. Left lungs were fixed in 10% formalin, paraffin embedded, sectioned, and stained with hematoxylin and eosin (H&E). To quantify differences between lungs of mice infected with BCG and H37Rv(pLUC10-AS-SOD) at 17 mo, a pathologist (K.T.T.) assessed interstitial, peribronchiolar, alveolar, and perivascular infiltration, using a scoring system that assigned a number value of 0, 1, 2, 3, or 4 on the basis of the magnitude of inflammatory cell infiltration. Statistical significance was determined by Mann–Whitney nonparametric analysis. Apoptosis was evaluated by TUNEL (TdT [terminal deoxynucleotidyltransferase]-mediated dUTP nick end labeling) assessment of DNA fragmentation (R&D Diagnostics, Madison, WI), with percent values determined by averaging counts of two examiners from three representative fields.

RESULTS

ately posttransformation, during serial passage on solid medium the SOD-diminished strains grew almost as well as non-SODdiminished M. tuberculosis strains, approximating 80% the rate of the parent strain. When cultivated in liquid medium, the growth rates of the SOD-diminished strains varied on the basis of whether Tween 80 or glycerol was added, growing at 30% and 50% of the rate of the parent strain, respectively. SOD-diminished H37Rv Is Attenuated in Mice and Induces Interstitial Mononuclear Cell Infiltration

Compared with control strains, SOD-diminished strains were markedly attenuated in C57BL/6 (H-2b) mice (Figure 2). Although much of the decline between Day 0 and Day 1 values in Figure 2 reflects tissue distribution of the intravenously administered inoculum that characteristically approximates 90% to the liver, 10% to the spleen, and 1% to 5% to the lungs unless excessive clumping is present (23), SOD-attenuated bacilli exhibited lower counts the day after inoculation than expected from tissue distribution alone. This was followed by modest growth up to Day 14 and then a decline by Day 28. In contrast, organ counts of the control strains rose steadily between Days 1 and 28, and some mice died in Weeks 3 and 4. By Day 28 there was nearly a 100,000-fold difference between counts of the SOD-attenuated versus control strains in the lungs and the spleens. Lung infection with the SOD-attenuated H37Rv strains and their virulent controls produced markedly different histo-

In Vitro Characteristics of SOD-attenuated Mutants of H37Rv

To construct plasmids that achieve a partial reduction in SOD production by M. tuberculosis, we ligated a sodA DNA fragment into pHV203 (Figure 1A) and pLUC10, yielding pHV203-ASSOD and pLUC10-AS-SOD, respectively. To evaluate the expression of AS-sodA RNA in mycobacteria we used M. smegmatis instead of M. tuberculosis, as previous studies have shown that the 65-kDa heat shock protein promoter is recognized by both species (22). This showed strong expression of AS-sodA RNA in a mycobacterial host (Figure 1B). Next, the AS-sodA plasmids were electroporated into the virulent M. tuberculosis strain H37Rv, yielding mutants exhibiting diminished production of SOD (Figure 1C). The SOD-attenuated H37Rv strain was more susceptible to killing by hydrogen peroxide than its isogenic control without the AS-sodA DNA fragment (Figure 1D). Of note, M. tuberculosis tranformants containing pLUC10AS-SOD or pHV203-AS-SOD grew slowly after electroporation or when subcultured from freezer stocks, appearing as pinpoint colonies on agar at 4 wk and taking 7 to 8 wk before they reached a colony size comparable to 3- to 4-wk growth of the parent and control strains. However, despite this slow initial growth immedi-

Figure 2. Survival of M. tuberculosis H37Rv and transformants in vivo. Solid symbols indicate controls and open symbols represent SOD-diminished isolates: H37Rv (solid diamonds), H37Rv(pLUC10) (solid circles), H37Rv(pHV203) (solid squares), H37Rv(pLUC10-AS-SOD) (open circles) with results from two experiments displayed, and H37Rv(pHV203AS-SOD) (open squares). Initial inocula administered intravenously to each group of mice are shown as Day 0 counts. Day 1, 7, 14, and 28 values represent CFU from organs, displayed as mean values SEM. Six mice in each group at each time point produced a single data point, except for the Day 28 value for H37Rv(pLUC10), which had only a single survivor (not shown).

Edwards, Cynamon, Voladri, et al.: SOD-attenuated M. tuberculosis

2215

Figure 3. Infiltration of inflammatory cells over time in mouse lungs infected with SOD-diminished H37Rv(pLUC10-AS-SOD), H37Rv(pHV203-AS-SOD), and virulent H37Rv(pLUC10). Representative H&E-stained views were photographed with a 10 microscope objective for enlargement. Mice infected with H37Rv(pLUC10) did not survive to Day 28.

pathologic findings (Figure 3). Especially notable was the greater interstitial infiltration with mononuclear cells, mostly macrophages, at 24 h in mice infected with SOD-attenuated H37Rv, despite the smaller numbers of viable bacilli at this time (Figure 2). Furthermore, the microscopic features evolved differently over time for the two groups. The early interstitial mononuclear cell infiltrate diminished over the first 2 wk in mice infected with SOD-attenuated H37Rv, and again became prominent by Day 28, likely because of the development of a

cellular immune response in Weeks 3 and 4 postinfection. In contrast, lungs of mice infected with virulent H37Rv control strains showed minimal interstitial infiltration at 24 h, with a modest increase in septal thickening by Day 7. However, by Day 14 there was marked interstitial mononuclear cell infiltration as well as alveolar polymorphonuclear and mononuclear cell infiltration with prominent necrosis. If the mouse survived, the infiltration and necrosis were even more intense on Day 28.

Figure 4. Apoptosis in mouse lungs infected with SOD-diminished H37Rv(pLUC10-AS-SOD) and its virulent control, H37Rv(pLUC10). Representative TUNEL-stained views were photographed with a 40 microscope objective for enlargement. Nuclei of cells undergoing apoptosis stain dark blue with TUNEL, whereas light blue staining of extracellular amorphous material represents background [e.g., Day 14, H37Rv (pLUC10)]. The graph plots the percentage of cells that appear apoptotic over time as determined from review of multiple 40 magnification fields: H37Rv(pLUC10) (open columns) and H37Rv(pLUC10-AS-SOD) (gray columns). Mice infected with H37Rv(pLUC10) did not survive to Day 28 (*).

2216


SOD-attenuated H37Rv and Mononuclear Cell Apoptosis

TUNEL assessment of DNA fragmentation in the lungs of mice infected with SOD-diminished H37Rv showed that about 30% of the interstitial mononuclear cells were apoptotic on Day 1, Day 7, and Day 14 (Figure 4). Apoptosis increased markedly to involve 60% of total cells at 28 d, coinciding with the renewed interstitial mononuclear cell infiltration at this time (Figure 3). In contrast, although apoptosis was identified on Day 1 and Day 7 in mice infected with virulent M. tuberculosis, it appeared to involve primarily cells lining the alveolar spaces and only rare apoptotic mononuclear cells were observed in the interstitial space. By Day 14, mice infected with virulent bacilli exhibited minimal apoptosis ( 5%), despite the development of marked interstitial and alveolar infiltration. Apoptosis of mononuclear cells in the interstitial inflammatory infiltrate occurred almost exclusively in the lungs of mice infected with SOD-attenuated H37Rv. Mice Tolerate Prolonged Infection with SOD-attenuated H37Rv Better than with BCG

Next, the effect of prolonged infection with SOD-diminished H37Rv versus BCG was compared after tail vein administration of inocula of 3.8 106 CFU. After initial greater clearance of SOD-attenuated H37Rv in the first 24 h, the rates of decline in counts of viable bacilli in the lungs were comparable, so that an approximately 10-fold difference between strains was maintained over the 17 mo of observation (Figure 5A). In contrast, the SOD-attenuated strain was cleared more rapidly than BCG from spleens, and beyond 8 wk a 2 to 4 log-fold difference in counts of viable bacilli was evident between the strains. At 17 mo, counts of viable bacilli in the spleens were below the lower limits of detection (10 CFU) in 5 of 12 mice that had received H37Rv(pLUC10-AS-SOD) compared with none of 11 surviving BCG-infected mice. Mice infected with H37Rv(pLUC10-AS-SOD) gained more weight than mice infected with BCG (Figure 5B) and at 17 mo were 6.3 g heavier (34.1 4.6 vs. 27.8 4.9 g, mean value SD, p 0.01, twosample t test). At 17 mo postinoculation, nodular lesions and apical consolidation were evident in the lungs of BCG-infected mice (Figure 6). In contrast, the lungs of mice infected with SODattenuated H37Rv were more homogeneous. Scoring of the magnitude of inflammatory cell infiltration into various regions of the lung by a pathologist unaware of which strain each mouse had received indicated mild interstitial and peribronchiolar infiltration that was similar for both groups of mice. Lungs from BCG-infected mice exhibited significantly greater alveolar infiltration (p 0.01, Mann–Whitney), corresponding to the nodular lesions and apical areas of consolidation observed with the naked eye (Figure 6). The lungs from mice infected with SOD-attenuated H37Rv also showed significantly larger perivascular collections of lymphocytes, macrophages, and plasma cells (p 0.01, Mann–Whitney) as shown in Figure 6.

VOL 164

2001

which were prominent in mice infected with the SOD-attenuated strains but reduced in mice infected with virulent H37Rv as well as BCG (BCG data not shown). Compared with the clearance of BCG from mice, the SOD-attenuated mutant was eradicated more rapidly, particularly from the spleen. Finally, compared with BCG, the SOD-attenuated strain was better tolerated by the mice, as evidenced by superior weight gain and the appearance of the lungs at 17 mo postinoculation. The construction and testing of paired isolates of bacteria that differ only in a single gene product is an invaluable tool in unraveling the molecular pathogenesis of infection. Although allelic inactivation is now feasible in M. tuberculosis (28–30), it can be used only with genes that are not essential for in vitro growth, making it necessary to apply other techniques to investigations involving essential microbial genes. Antisense RNA methods were originally described as a means of specifically diminishing the production of microbial factors (31). Although now more commonly employed in eukaryotic cells, where the presence of two copies of the chromosome make allelic inactivation impractical, endogenous antisense RNA expression has been used to attenuate a virulent strain of Staphylococcus aureus (32) and applied to mycobacteria, including M. smegmatis and M. bovis (33, 34). The previously described inability to inactivate M. tuberculosis sodA (18) and slow growth of our SOD-attenuated H37Rv strains suggest that SOD is either essential for M. tuberculosis viability or that sodA deletion mutants would grow too poorly to be recovered in conventional laboratory media. Thus, although endogenous antisense RNA expression may not be as rigorous a technique as allelic inactivation for determining the role in pathogenesis of specific microbial factors, it permits a partial reduction of a gene product under circumstances in which allelic inactivation prevents cultivation of the microbe. There are several plausible reasons for the marked attenuation of SOD-diminished M. tuberculosis. SOD is important in maintaining an optimal intracellular redox environment in the bacterial cytoplasm that is needed to prevent lipid peroxidation, DNA cross-linking, and cysteine–cysteine bonding of essential enzymes (35). Our mutants grew slowly in vitro under conditions that required cultivation from single or dispersed bacilli (e. g., postelectroporation or in broth containing Tween 80); however, growth was restored to a rate almost comparable to parent and control strains under conditions that allowed the organisms to accumulate (e.g., passage on agar). This difference may reflect the ability of other oxygen-scavenging moieties such as mannosylated lipoarabinomannan, phenolic glycolipid 1, and mycobacterial sulfatides to complement SOD deficiency and partially protect the microbe against oxygenrelated toxicity under conditions of cell aggregation. The slow

DISCUSSION In this investigation, we used antisense techniques to reduce SOD production by H37Rv, a well-characterized virulent M. tuberculosis strain (24). An SOD-attenuated mutant was more susceptible than a control strain to killing by hydrogen peroxide, similar to SOD-deficient mutants of other bacterial species (25–27). Also, SOD-attenuated mutants survived less well in a mouse model, exhibiting a fall in bacillary load in the spleens and lungs over time compared with a rapid rise by the virulent control strains. Falls in bacillary counts in the lungs correlated with interstitial mononuclear cell infiltration and apoptosis,

Figure 5. Infection of mice with BCG and SOD-diminished H37Rv. BCG Tice (solid circles), SOD-diminished H37Rv (open circles). Bacilli recovered from right lungs and spleens (A) displayed as means SEM. Each data point was derived from 6 mice except for the 17-mo values, which used 11 mice for BCG and 12 mice for H37Rv(pLUC10-AS-SOD). Weight of mice (B) displayed as means SEM.


2217

Figure 6. Left lungs of C57BL/ 6 mice at 17 mo postinoculation. Top: Whole lungs of mice that received BCG and SODdiminished H37Rv. Bottom: (A– F) H&E-stained longitudinal crosssections of the lungs with letters corresponding to those in the upper panels, enlarged and photographed through a 4 microscope objective. (G) Representative perivascular collection of lymphocytes, macrophages, and plasma cells from a mouse infected with SOD-diminished H37Rv; 40 microscope objective.

in vitro growth raises the possibility that the SOD-attenuated strains had a reduced growth rate in vivo independent of killing by host defenses. However, whereas the SOD-attenuated strains grew at 30% to 80% of the rate of SOD-proficient H37Rv control strains under different conditions in vitro, in animals they actually decreased in number over 28 d rather than just growing more slowly. As a result, the numbers of SOD-attenuated organisms recovered from the lungs and spleens at 28 d were nearly 100,000-fold fewer than those of H37Rv (Figure 2). These findings strongly suggest that the reduced growth of the SOD-attenuated organisms in vivo did not result primarily from a lower intrinsic capacity to replicate, but to interactions between the mycobacteria and host. A second factor affecting attenuation might be enhanced susceptibility to the reactive oxygen intermediates (ROIs) of phagocytes, a possibility supported by the susceptibility of the SOD-attenuated mutant to hydrogen peroxide in vitro. However, studies with virulent M. tuberculosis suggest that nitric oxide is more important than ROIs in the killing of mycobacteria (36). Furthermore, M. tuberculosis can enter macrophages via complement receptors, thereby avoiding an oxidative burst (37). Such data suggest that ROIs do not exert an important direct antimicrobial effect against pathogenic M. tuberculosis strains in vivo; however, they do not address their effect against the SOD-attenuated strains created in this study. Therefore, we focused on the marked difference in interstitial mononuclear cell infiltration that was unique to infection with SOD-attenuated H37Rv and considered whether the recruitment and apoptosis of mononuclear cells, which kills tubercle bacilli in vitro (38–40), might be a factor in the killing of SOD-attenuated H37Rv in vivo. In mice infected with SOD-attenuated H37Rv, apoptosis occurred among the interstitial mononuclear cell infiltrate, and the timing of infiltration and apoptosis of interstitial mononuclear cells correlated with falls in the bacillary counts. In contrast, minimal interstitial mononuclear cell apoptosis occurred in mice infected with virulent H37Rv, although apoptosis of alveolar cells was observed and possibly reflects lung injury from mycobacterial cord factor and other microbial factors similar to apoptosis induced by the lipopolysaccharide of gram-negative bacteria (41). Although the precise identity of the

apoptotic cells in the interstitial infiltrate was not determined, on the first day postinfection they likely represented monocytes and monocyte-derived macrophages. The types of cells undergoing apoptosis in the renewed interstitial infiltrate on Day 28 are less certain but may have included T lymphocytes along with cells of monocytic derivation. Furthermore, acquired immune responses involving T cells may have contributed to the increase in apoptosis observed at this time. So, in summary, we have observed marked differences in interstitial mononuclear cell infiltration, apoptosis, and pathogenicity exhibited by M. tuberculosis strains that are identical except for the amount of SOD they produce. Regarding the molecular mechanisms underlying these observations, it is reasonable to consider the possibility that the iron-cofactored SOD of M. tuberculosis contributes to infection pathogenesis by scavenging ROIs, thereby blocking both NF-B activation and apoptosis of mononuclear cells (42, 43). The activation of NF-B and apoptosis are redox-sensitive intracellular processes that are inhibited by mammalian SOD (44–46), and the iron-cofactored SOD of M. tuberculosis is exported in abundant amounts via a leader peptide-independent mechanism (15, 17). Investigations involving mycobacteria, ROIs, NF-B activation, and apoptosis are complex, with the observations dependent in part on the mycobacterial strain evaluated (e.g., the use of strain H37Ra favors NF-B activation because its lipoarabinomannan [LAM] is not mannosylated and cannot scavenge ROIs as efficiently as the mannosylated LAM of virulent M. tuberculosis strains), other experimental conditions, and an agonist effect of NF-B activation on the induction of mammalian SOD production (38, 47–51). However, emerging data suggest that the stimulation of Toll-like receptors by mycobacterial factors initiates intracellular signaling that culminates in both NF-B activation and apoptosis (52–54). Furthermore, attempts to discern the roles of different ROIs suggest that superoxide anion, which is the target of superoxide dismutase, is more effective than hydroxyl radicals or hydrogen peroxide as a catalyst for both processes (42, 55). On the basis of the ability of SOD to detoxify superoxide anion and its export in large amounts by virulent M. tuberculosis strains and BCG (15, 17, 56–58), we believe that in addition to any direct killing of SOD-attenuated strains by ROIs, our

2218


experimental results may be related to an effect of SOD reduction on two redox-sensitive host responses to infection. First, the mononuclear cell infiltration observed in lungs infected by SOD-attenuated strains may occur because NF-B activation in alveolar macrophages is unopposed by SOD, thereby facilitating the production of cytokines and chemokines, which attract additional inflammatory cells. Second, the SOD-attenuated strains induce apoptosis of mononuclear cells more readily, as exported SOD is not present to inhibit the process. Conversely, iron-cofactored SOD appears to affect the pathogenesis of tuberculosis via a combination of reduced early mononuclear cell infiltration and inhibition of apoptosis, thereby impairing the innate immune response to infection and allowing the organism to persist (59, 60). There is precedence among other intracellular pathogens for blocking macrophage apoptosis as a means to enable continued intracellular parasitism (61–64). Furthermore, the overexpression of human manganese-cofactored SOD in cancer cells suppresses both NF-B activation and apoptosis (65), establishing feasibility for the hypothesis that exported M. tuberculosis SOD might do the same. In conclusion, we have shown that reducing SOD production by M. tuberculosis enhances rapid mononuclear cell infiltration into the lung and apoptosis of these cells. This suggests that SOD contributes to the pathogenesis of tuberculosis by preventing the early elimination of M. tuberculosis by innate immune responses, an especially critical event for an inhaled pathogen that infects with an inoculum of only a few bacilli. Endogenous antisense RNA expression is a useful strategy for determining the role of essential microbial genes in infection pathogenesis, overcoming specific limitations of allele inactivation methods by enabling the construction of mutants with partial phenotypes. Finally, by reducing SOD we have attenuated M. tuberculosis so that a mammalian host tolerates it better than BCG during prolonged infection. Acknowledgment : The authors thank Jim Graham and Luc Van Kaer for review of the manuscript. E-293 antiserum was obtained from Colorado State University under NIH, NIAID contract N01 AI-75320 (TB Research Materials and Vaccine Testing). The authors thank Douglas Young for providing pBAK14, Richard Young for pY6002, Robert Cooksey for pLUC10, and William Jacobs, Jr. for M. smegmatis mc2155.

References 1. Institute of Medicine, Committee on the Elimination of Tuberculosis in the United States. Ed. Geiter L. Ending neglect: the elimination of tuberculosis in the United States. Washington, DC: National Academy Press; 2000. 2. Murray CJ, Salomon JA. Modeling the impact of global tuberculosis control strategies. Proc Natl Acad Sci USA 1998;95:13881–13886. 3. Berthet FX, Lagranderie M, Gounon P, Laurent-Winter C, Ensergueix D, Chavarot P, Thouron F, Maranghi E, Pelicic V, Portnoi D, et al. Attenuation of virulence by disruption of the Mycobacterium tuberculosis erp gene. Science 1998;282:759–762. 4. Chen P, Ruiz RE, Li Q, Silver RF, Bishai WR. Construction and characterization of a Mycobacterium tuberculosis mutant lacking the alternate sigma factor gene, sigF. Infect Immun 2000;68:5575–5580. 5. Glickman MS, Cox JS, Jacobs WR Jr. A novel mycolic acid cyclopropane synthetase is required for coding, persistence, and virulence of Mycobacterium tuberculosis. Mol Cell 2000;5:717–727. 6. McKinney JD, Honer zu Bentrup K, Munoz-Elias EJ, Miczak A, Chen B, Chan WT, Swenson D, Sacchettini JC, Jacobs WR Jr, Russell DG. Persistence of Mycobacterium tuberculosis in macrophages and mice requires the glyoxylate shunt enzyme isocitrate lyase. Nature 2000;406:735–738. 7. Guleria I, Teitelbaum R, McAdam RA, Kalpana G, Jacobs WR Jr, Bloom BR. Auxotrophic vaccines for tuberculosis. Nat Med 1996;2: 334–337. 8. Hondalus MK, Bardarov S, Russell R, Chan J, Jacobs WR Jr, Bloom BR. Attenuation of and protection induced by a leucine auxotroph of Mycobacterium tuberculosis. Infect Immun 2000;68:2888–2898. 9. Jackson M, Phalen SW, Lagranderie M, Ensergueix D, Chavarot P, Marchal G, McMurray DN, Gicquel B, Guilhot C. Persistence and protec-

10.

11.

12.

13.

14.

15.

16.

17.

18.

19.

20.

21.

22.

23.

24.

25.

26.

27.

28.

29.

30.

VOL 164

2001

tive efficacy of a Mycobacterium tuberculosis auxotroph vaccine. Infect Immun 1999;67:2867–2873. Smith DA, Parish T, Stoker NG, Bancroft GJ. Characterization of auxotrophic mutants of Mycobacterium tuberculosis and their potential as vaccine candidates. Infect Immun 2001;69:1142–1150. Fields PI, Swanson RV, Haidaris CG, Heffron F. Mutants of Salmonella typhimurium that cannot survive within the macrophage are avirulent. Proc Natl Acad Sci USA 1986;83:5189–5193. Fang FC, DeGroote MA, Foster JW, Baumler AJ, Ochsner U, Testerman T, Bearson S, Giard JC, Xu Y, Campbell G, Laessig T. Virulent Salmonella typhimurium has two periplasmic Cu, Zn-superoxide dismutases. Proc Natl Acad Sci USA 1999;96:7502–7507. Igwe EI, Russmann H, Roggenkamp A, Noll A, Autenrieth IB, Heesemann J. Rational live oral carrier vaccine design by mutating virulence-associated genes of Yersinia enterocolitica. Infect Immun 1999; 67:5500–5507. Sadosky AB, Wilson JW, Steinman HM, Shuman HA. The iron superoxide dismutase of Legionella pneumophila is essential for viability. J Bacteriol 1994;176:3790–3799. Zhang Y, Lathigra R, Garbe T, Catty D, Young D. Genetic analysis of superoxide dismutase, the 23 kilodalton antigen of Mycobacterium tuberculosis. Mol Microbiol 1991;5:381–391. Piddington DL, Fang FC, Laessig T, Cooper AM, Orme IM, Buchmeier NA. Cu,Zn superoxide dismutase of Mycobacterium tuberculosis contributes to survival in activated macrophages that are generating an oxidative burst. Infect Immun 2001;69:4980–4987. Harth G, Horwitz MA. Export of recombinant Mycobacterium tuberculosis superoxide dismutase is dependent upon both information in the protein and mycobacterial export machinery. A model for studying export of leaderless proteins by pathogenic mycobacteria. J Biol Chem 1999;274:4281–4292. Dussurget O, Stewart G, Neyrolles O, Pescher P, Young D, Marchal G. Role of Mycobacterium tuberculosis copper–zinc superoxide dismutase. Infect Immun 2001;69:529–533. Aldovini A, Husson RN, Young RA. The uraA locus and homologous recombination in Mycobacterium bovis BCG. J Bacteriol 1993;175:7282– 7289. Cooksey RC, Crawford JT, Jacobs WR Jr, Shinnick TM. A rapid method for screening antimicrobial agents for activities against a strain of Mycobacterium tuberculosis expressing firefly luciferase. Antimicrob Agents Chemother 1993;37:1348–1352. Lakey DL, Voladri RK, Edwards KM, Hager C, Samten B, Wallis RS, Barnes PF, Kernodle DS. Enhanced production of recombinant Mycobacterium tuberculosis antigens in Escherichia coli by replacement of low-usage codons. Infect Immun 2000;68:233–238. Dhandayuthapani S, Via LE, Thomas CA, Horowitz PM, Deretic D, Deretic V. Green fluorescent protein as a marker for gene expression and cell biology of mycobacterial interactions with macrophages. Mol Microbiol 1995;17:901–912. Orme IM, Collins FM. Mouse model of tuberculosis. In: Bloom BR, editor. Tuberculosis; pathogenesis, protection, and control. 1st edition. Washington, DC: ASM Press; 1994, p 113–134. Cole ST, Brosch R, Parkhill J, Garnier T, Churcher C, Harris D, Gordon SV, Eiglmeier K, Gas S, Barry CE III, et al. Deciphering the biology of Mycobacterium tuberculosis from the complete genome sequence. Nature 1998;393:537–544. Carlioz A, Touati D. Isolation of superoxide dismutase mutants in Escherichia coli: is superoxide dismutase necessary for aerobic life? EMBO J 1986;5:623–630. Hassett DJ, Schweizer HP, Ohman DE. Pseudomonas aeruginosa sodA and sodB mutants defective in manganese- and iron-cofactored superoxide dismutase activity demonstrate the importance of the ironcofactored form in aerobic metabolism. J Bacteriol 1995;177:6330– 6337. San Mateo LR, Toffer KL, Kawula TH. The sodA gene of Haemophilus ducreyi encodes a hydrogen peroxide-inhibitable superoxide dismutase. Gene 1998;207:251–257. Bardarov S, Kriakov J, Carriere C, Yu S, Vaamonde C, McAdam RA, Bloom BR, Hatfull GF, Jacobs WR Jr. Conditionally replicating mycobacteriophages: a system for transposon delivery to Mycobacterium tuberculosis. Proc Natl Acad Sci USA 1997;94:10961–10966. Pelicic V, Jackson M, Reyrat JM, Jacobs WR Jr, Gicquel B, Guilhot C. Efficient allelic exchange and transposon mutagenesis in Mycobacterium tuberculosis. Proc Natl Acad Sci USA 1997;94:10955–10960. Parish T, Stoker NG. Use of a flexible cassette method to generate a double unmarked Mycobacterium tuberculosis tlyA plcABC mutant by gene replacement. Microbiology 2000;146:1969–1975.


2219

31. Coleman J, Green PJ, Inouye M. The use of RNAs complementary to specific mRNAs to regulate the expression of individual bacterial genes. Cell 1984;37:429–436. 32. Kernodle DS, Voladri RK, Menzies BE, Hager CC, Edwards KM. Expression of an antisense hla fragment in Staphylococcus aureus reduces toxin production in vitro and attenuates lethal activity in a murine model. Infect Immun 1997;65:179–184. 33. Parish T, Stoker NG. Development and use of a conditional antisense mutagenesis system in mycobacteria. FEMS Microbiol Lett 1997;154: 151–157. 34. Wilson T, de Lisle GW, Marcinkeviciene JA, Blanchard JS, Collins DM. Antisense RNA to ahpC, an oxidative stress defence gene involved in isoniazid resistance, indicates that AhpC of Mycobacterium bovis has virulence properties. Microbiology 1998;144:2687–2695. 35. Powis G, Briehl M, Oblong J. Redox signaling and the control of cell growth and death. Pharmacol Ther 1995;68:149–173. 36. Chan J, Xing Y, Magliozzo RS, Bloom BR. Killing of virulent Mycobacterium tuberculosis by reactive nitrogen intermediates produced by activated murine macrophages. J Exp Med 1992;175:1111–1122. 37. Schlesinger LS. Entry of Mycobacterium tuberculosis into mononuclear phagocytes. Curr Top Microbiol Immunol 1996;215:71–96. 38. Oddo M, Renno T, Attinger A, Bakker T, MacDonald HR, Meylan PR. Fas ligand-induced apoptosis of infected human macrophages reduces the viability of intracellular Mycobacterium tuberculosis. J Immunol 1998;160:5448–5454. 39. Duan L, Gan H, Arm J, Remold HG. Cytosolic phospholipase A2 participates with TNF-alpha in the induction of apoptosis of human macrophages infected with Mycobacterium tuberculosis H37Ra. J Immunol 2001;166:7469–7476. 40. Molloy A, Laochumroonvorapong P, Kaplan G. Apoptosis, but not necrosis, of infected monocytes is coupled with killing of intracellular bacillus Calmette-Guerin. J Exp Med 1994;180:1499–1509. 41. Kitamura Y, Hashimoto S, Mizuta N, Kobayashi A, Kooguchi K, Fujiwara I, Nakajima H. Fas/FasL-dependent apoptosis of alveolar cells after lipopolysaccharide-induced lung injury in mice. Am J Respir Crit Care Med 2001;163:762–769. 42. Wang S, Leonard SS, Castranova V, Vallyathan V, Shi X. The role of superoxide radical in TNF- induced NF-B activation. Ann Clin Lab Sci 1999;29:192–199. 43. Jabs T. Reactive oxygen intermediates as mediators of programmed cell death in plants and animals. Biochem Pharmacol 1999;57:231–245. 44. Wheeler MD, Yamashina S, Froh M, Rusyn I, Thurman RG. Adenoviral gene delivery can inactivate Kupffer cells: role of oxidants in NF-B activation and cytokine production. J Leukoc Biol 2001;69:622–630. 45. Brockhaus F, Brune B. Overexpression of CuZn superoxide dismutase protects RAW 264.7 macrophages against nitric oxide cytotoxicity. Biochem J 1999;338:295–303. 46. Kuroda M, Himei K, St Clair DK, Urano M, Yoshino T, Akagi T, Asaumi J, Akaki S, Takeda Y, Kanazawa S, Hiraki Y. Overexpression of manganese superoxide dismutase gene suppresses spontaneous apoptosis without a resultant alteration in in vivo growth of the mouse fibrosarcoma, FSa-II. Anticancer Res 2000;20:7–10. 47. Tchou-Wong KM, Tanabe O, Chi C, Yie TA, Rom WN. Activation of NF-B in Mycobacterium tuberculosis-induced interleukin-2 receptor expression in mononuclear phagocytes. Am J Respir Crit Care Med 1999;159:1323–1329. 48. Durrbaum-Landmann I, Gercken J, Flad HD, Ernst M. Effect of in vitro infection of human monocytes with low numbers of Mycobacterium tuberculosis bacteria on monocyte apoptosis. Infect Immun 1996;64: 5384–5389.

49. Brown MC, Taffet SM. Lipoarabinomannans derived from different strains of Mycobacterium tuberculosis differentially stimulate the activation of NF-B and KBF1 in murine macrophages. Infect Immun 1995;63:1960–1968. 50. Jones BE, Lo CR, Liu H, Pradhan Z, Garcia L, Srinivasan A, Valentino KL, Czaja MJ. Role of caspases and NF-B signaling in hydrogen peroxide- and superoxide-induced hepatocyte apoptosis. Am J Physiol Gastrointest Liver Physiol 2000;278:G693–G699. 51. Mattson MP, Goodman Y, Luo H, Fu W, Furukawa K. Activation of NF-B protects hippocampal neurons against oxidative stress-induced apoptosis: evidence for induction of manganese superoxide dismutase and suppression of peroxynitrite production and protein tyrosine nitration. J Neurosci Res 1997;49:681–697. 52. Means TK, Wang S, Lien E, Yoshimura A, Golenbock DT, Fenton MJ. Human toll-like receptors mediate cellular activation by Mycobacterium tuberculosis. J Immunol 1999;163:3920–3927. 53. Means TK, Jones BW, Schromm AB, Shurtleff BA, Smith JA, Keane J, Golenbock DT, Vogel SN, Fenton MJ. Differential effects of a tolllike receptor antagonist on Mycobacterium tuberculosis-induced macrophage responses. J Immunol 2001;166:4074–4082. 54. Aliprantis AO, Yang RB, Weiss DS, Godowski P, Zychlinsky A. The apoptotic signaling pathway activated by Toll-like receptor-2. EMBO J 2000;19:3325–3336. 55. Shrivastava A, Aggarwal BB. Antioxidants differentially regulate activation of nuclear factor-B, activator protein-1, c-Jun amino-terminal kinases, and apoptosis induced by tumor necrosis factor: evidence that JNK and NF-B activation are not linked to apoptosis. Antioxid Redox Signal 1999;1:181–191. 56. Deshpande RG, Khan MB, Bhat DA, Navalkar RG. Superoxide dismutase activity of Mycobacterium tuberculosis isolated from tuberculosis patients and the immunoreactivity of superoxide dismutase from M. tuberculosis H37Rv. Tuber Lung Dis 1993;74:388–394. 57. Raynaud C, Etienne G, Peyron P, Laneelle MA, Daffe M. Extracellular enzyme activities potentially involved in the pathogenicity of Mycobacterium tuberculosis. Microbiology 1998;6:71–78. 58. Florio W, Freer G, Daila CB, Batoni G, Maisetta G, Senesi S, Campa M. Comparative analysis of subcellular distribution of protein antigens in Mycobacterium bovis bacillus Calmette-Guerin. Can J Microbiol 1997; 43:744–750. 59. Kornfeld H, Mancino G, Colizzi V. The role of macrophage cell death in tuberculosis. Cell Death Differ 1999;6:71–78. 60. Keane J, Remold HG, Kornfeld H. Virulent Mycobacterium tuberculosis strains evade apoptosis of infected alveolar macrophages. J Immunol 2000;164:2016–2020. 61. Nash PB, Purner MB, Leon RP, Clarke P, Duke RC, Curiel TJ. Toxoplasma gondii-infected cells are resistant to multiple inducers of apoptosis. J Immunol 1998;160:1824–1830. 62. Moore KJ, Matlashewski G. Intracellular infection by Leishmania donovani inhibits macrophage apoptosis. J Immunol 1994;152:2930–2937. 63. Gao L, Abu KY. Hijacking of apoptotic pathways by bacterial pathogens. Microbes Infect 2000;2:1705–1719. 64. Fan T, Lu H, Hu H, Shi L, McClarty GA, Nance DM, Greenberg AH, Zhong G. Inhibition of apoptosis in chlamydia-infected cells: blockade of mitochondrial cytochrome c release and caspase activation. J Exp Med 1998;187:487–496. 65. Manna SK, Zhang HJ, Yan T, Oberley LW, Aggarwal BB. Overexpression of manganese superoxide dismutase suppresses tumor necrosis factor-induced apoptosis and activation of nuclear transcription factor-B and activated protein-1. J Biol Chem 1998;273: 13245–13254.