Protection Elicited by Two Glutamine Auxotrophs of Mycobacterium tuberculosis and In Vivo Growth Phenotypes of the Four Unique Glutamine Synthetase Mutants in a Murine Model

Abstract

We generated four individual glutamine synthetase (GS) mutants (ΔglnA1, ΔglnA2, ΔglnA3, and ΔglnA4) and one triple mutant (ΔglnA1EA2) of Mycobacterium tuberculosis to investigate the roles of GS enzymes. Subcutaneous immunization with the ΔglnA1EA2 and ΔglnA1 glutamine auxotrophic mutants conferred protection on C57BL/6 mice against an aerosol challenge with virulent M. tuberculosis, which was comparable to that provided by Mycobacterium bovis BCG vaccination.

Mycobacterium tuberculosis, the causative agent of human tuberculosis (TB), infects one-third of the world's population and is responsible for 3 million deaths annually (4). In recent years, the convergence of the TB and human immunodeficiency virus epidemics, as well as the emergence of multidrug-resistant M. tuberculosis strains, has magnified this global public health tragedy and has accelerated efforts to develop improved interventions against tuberculosis, including new TB vaccines (3, 6, 16). Among the most immunogenic novel vaccines are live, attenuated M. tuberculosis strains. Recent advances in tools to genetically modify M. tuberculosis strains have allowed the creation of defined mutants that are highly attenuated and yet still undergo limited replication in mice. Studies in immunocompromised and immunocompetent mice have demonstrated that safe, attenuated M. tuberculosis vaccine strains with multiple gene deletions that effectively protect against airborne infection with M. tuberculosis can be generated (22). As additional gene deletion targets are identified, more unique attenuated live vaccines can be produced and tested in animal models.

Potential gene deletion targets for attenuated vaccines are the genes encoding the glutamine synthetase (GS) enzymes, which are important proteins involved in prokaryotic nitrogen metabolism. GS converts glutamate and ammonia to glutamine, and glutamate synthetase transfers the amide group from glutamine to 2-ketoglutarate to produce two glutamate molecules. So far, four different GSs have been identified based on the number of subunits, heat stability, and covalent posttranslational modification (13). In enteric bacteria, a single glnA gene encodes a GS type I enzyme, and a glnA null mutant is a glutamine auxotroph; however, other bacteria have two or three different types of GSs (5, 13, 26, 27). Interestingly, the M. tuberculosis genome encodes four GS homologues. GlnA1, GlnA3, and GlnA4 catalyze the synthesis of l-glutamate, and GlnA2 catalyzes the synthesis of d-glutamine and d-isoglutamine, which are required for generating the mycobacterial peptidoglycan layer. Recently, Tullius et al. have shown that inactivation of the M. tuberculosis glnA1 gene creates a glutamine auxotroph and that glnA1 is essential for growth in both differentiated THP-1 cells and guinea pigs (25). Although the other three GS genes were apparently transcribed, these glutamine synthetase gene products of M. tuberculosis did not complement the loss of the GlnA1 protein. Surprisingly, the GlnA2, GlnA3, and GlnA4 proteins are not essential for in vitro growth (7). However, the in vivo growth requirements for these M. tuberculosis gln mutants have not been evaluated.

In some bacteria, GS is regulated via adenylylation of the subunits within the aggregate; catalytic activity is inversely proportional to the extent of adenylylation. The adenylylation and deadenylylation of a tyrosine side chain on each GS subunit are catalyzed by adenylyl transferase (ATase), encoded by the glnE gene (12-14, 20). The ATase reaction is highly specific for GS. The importance of ATase in M. tuberculosis biology has been suggested by genetic disruption of the glnE gene, which was lethal (17, 18).

For these studies, we hypothesized that the different M. tuberculosis GS proteins might function under different environmental conditions, at different stages of pathogenesis, and/or for in vivo growth. In this paper, we examine the roles of the TB GS proteins in a mouse model under various in vitro conditions. Since the identification of additional gene deletion targets is essential for the further development of nonreplicating mutant strains of M. tuberculosis as vaccine candidates (9, 11, 19, 22-24), we also evaluated whether immunization with two glutamine auxotrophs, ΔglnA1 and ΔglnA1EA2, would induce protective immunity in a mouse model of pulmonary tuberculosis.

Analyses of M. tuberculosis glnA1, glnA2, glnA3, glnA4, and glnA1EA2 mutants.

To study the possible functions of the gln genes, we generated individual gln gene mutants by a specialized transduction method (1). The resulting transductants were verified for the generation of the proper mutations by Southern blots of total DNAs isolated from independent mutants (Fig. 1A; see Fig. S1 in the supplemental material).

FIG. 1.

Southern blot analysis of wild-type H37Rv and five gln mutants (ΔglnA1, ΔglnA2, ΔglnA3, ΔglnA4, and ΔglnA1EA2). (A and B) Genomic DNAs from the wild type and mutants were isolated, digested, and probed with regions flanking the M. tuberculosis gln genes. The double arrows (⇒) indicate the locations of the genomic fragments of Δgln mutants that hybridized, and the single arrows (←) indicate the fragments of H37Rv genomic DNA. (B) Maps of the glnA1EA2 genomic regions in the H37Rv and ΔglnA1EA2 mutants. The restriction sites and probe locations (for upstream of glnA2, for glnA1, and for glnE) are indicated. P, PstI; S, StyI.

A surprising result of this study was the ability to generate a ΔglnA1EA2 mutant, which was a glutamine auxotroph (Fig. 1B). To verify the deletion of three gln genes in the ΔglnA1EA2 mutant, we probed the genomic DNAs from H37Rv and two independent clones of ΔglnA1EA2 with a downstream region flanking the glnA2 (Fig. 1B, a), glnA1 (Fig. 1B, b), and glnE (Fig. 1B, c) fragments. Both the glnA1 and glnE probes did not hybridize the ΔglnA1EA2 genomic DNA, which confirmed that those two clones are true triple mutants. Disruption of the glnE gene of M. tuberculosis yielded a lethal phenotype, suggesting that GS adenylylation is a critical process in mycobacteria (17, 18). Since GlnE regulates GlnA1 activity, the lethality of the ΔglnE mutant is likely due to depletion of the intracellular pool of glutamate in the mutant. However, it was reported that the medium used for plating transductants contained a significant amount of glutamate (3 mM) (17, 18). We also attempted to make a ΔglnE mutant, but failed. Generation of the ΔglnA1EA2 mutant might be explained by the fact that no glutamine synthetase (GlnA1) was expressed in the ΔglnA1EA2 mutant, and therefore, the glutamate was not depleted in the mutant.

To further evaluate the importance of the GS genes in M. tuberculosis, we generated single mutants in the glnA1 to glnA4 genes. Consistent with the findings of Harth et al. (7), a deletion in the GlnA1 gene yielded a glutamine auxotroph. In contrast, the ΔglnA2, ΔglnA3, and ΔglnA4 mutants were not glutamine auxotrophs and had no growth defects under the conditions tested (microaerobic and aerobic conditions, different nitrogen sources, different pHs, and high-salt conditions [data not shown]). Complemented strains were generated by transforming the mutant strains with an integrating mycobacterium plasmid containing the M. tuberculosis glnA1 gene (ΔglnA1-C-TB) or the Mycobacterium smegmatis glnA1 (ΔglnA1-C-MS) gene under its own promoter and under the hsp60 promoter, respectively. Both complemented strains were glutamine prototrophs.

In vivo growth and survival.

The attenuation of the Δgln mutants was assessed by infection of immunocompromised SCID mice and immunocompetent BALB/c mice. SCID mice that were infected intravenously with 10⁶ CFU of the M. tuberculosis H37Rv, ΔglnA2, ΔglnA3, and ΔglnA4 strains and two complemented strains rapidly succumbed to the resulting infection: the mean survival time (MST) was 25.5 days for H37RV, 26 days for the ΔglnA2 strain, 21 days for the ΔglnA3 and ΔglnA4 strains, 22 days for ΔglnA1-C-TB, and 42 days for ΔglnA1-C-MS. By contrast, all mice infected with the ΔglnA1 and ΔglnA1EA2 mutants survived for more than 71 weeks (MST, 500 days), at which point the experiment was terminated (Fig. 2A). Enumeration of bacterial loads in SCID mice infected with the H37Rv, ΔglnA2, ΔglnA3, and ΔglnA4 strains and two complemented strains showed a rapid increase in bacterial numbers in the spleen, liver, and lung. Interestingly, mice infected with the ΔglnA1 and ΔglnA1EA2 mutants showed a sharp drop in bacterial numbers in the spleen, liver, and lung by day 21 (data not shown).

FIG. 2.

In vivo growth of H37Rv and Δgln mutants in immunocompetent (C57BL/6) and immunocompromised (SCID) mice. The survival of immunocompromised mice (A) infected with H37Rv and various Δgln mutants and of immunocompetent mice (B) showed that the mice infected with the ΔglnA2, ΔglnA3, and ΔglnA4 mutants and the complemented strains by the intravenous route had lengths of survival similar to that of the wild type. (C) The bacterial counts of the ΔglnA2, ΔglnA3, and ΔglnA4 mutants and the complemented strains in spleen, liver, and lung were determined in the immunocompetent mice, and they were similar to those of H37Rv-infected mice. (D) Histopathology of the lungs of mice infected with the Δgln mutants, two complementing strains, and the wild type examined at 16 weeks postinfection.

Survival studies showed that BALB/c mice infected with the H37Rv, ΔglnA2, ΔglnA3, ΔglnA4, and ΔglnA1-C-TB strains succumbed to death by day 203 (average, 162 days), while mice infected with an identical dose of the ΔglnA1-C-MS strain succumbed to death between days 215 and 313 (average, 244 days). The longer mean survival period for the ΔglnA1-C-MS-infected mice suggests that M. tuberculosis GS may have other important functions in the pathogenesis of M. tuberculosis, which the M. smegmatis GS cannot complement. Mice infected with the ΔglnA1 and ΔglnA1EA2 mutants survived longer than 600 days, at which point the experiment was terminated (Fig. 2B).

At 3 weeks after infection, the lungs of BALB/c mice infected with the ΔglnA1 and ΔglnA1EA2 mutants showed no detectable bacterial CFU; these mutants were found in the spleen and liver at 3 weeks postinfection, but the number of organisms in these organs declined over the next 8 weeks. The numbers of bacterial CFU in the lungs, spleens, and livers of mice infected with other gln mutants were similar to those of mice infected with H37Rv at 3, 8, and 16 weeks (Fig. 2C).

Histopathological examination of the lungs from immunocompetent mice infected with the H37Rv, ΔglnA2, ΔglnA3, and ΔglnA4 strains and two complemented strains showed severe, diffuse lobar granulomatous pneumonia at 16 weeks (Fig. 2D). The pneumonia affected more than half of the lung, and the affected areas of the lung were severely consolidated, with loss of air spaces. No pathological changes were observed in the mice infected with the ΔglnA1 and ΔglnA1EA2 mutants. These two mutants showed typical auxotrophic mutant phenotypes in mice.

Vaccination studies.

Since recent studies have demonstrated the potential of attenuated mutant strains of M. tuberculosis as TB vaccines (2, 8, 9, 19, 21, 23, 24), we tested the capacities of the two glutamine auxotrophs (ΔglnA1 and ΔglnA1EA2 mutants) to protect against a virulent aerogenic challenge in a mouse model of pulmonary tuberculosis. For this study, the mice were vaccinated subcutaneously with 10⁶ CFU of either the ΔglnA1 or ΔglnA1EA2 mutant and then challenged by aerosol with 200 CFU of M. tuberculosis Erdman. At the time of challenge, the ΔglnA1 and ΔglnA1EA2 mutants could not be detected in the spleens or lungs of the vaccinated mice. In the naive control, the bacterial CFU values were increased by 10⁵-fold in the lung during the first month after challenge. Similarly, substantial dissemination and growth in the spleen were detected within 1 month of challenge in naive controls. In contrast, mice immunized with a single dose of the ΔglnA1 or ΔglnA1EA2 mutant showed statistically significant reductions (P < 0.05) in lung and spleen CFU values relative to naive controls. Mice vaccinated with Mycobacterium bovis BCG showed similar reductions in organ bacterial burdens compared with the nonimmunized controls (Fig. 3A). In these aerogenic-challenge studies, no significant differences were detected between the lung and spleen CFU values for mice vaccinated with the ΔglnA1 strain, the ΔglnA1EA2 mutant, or BCG.

FIG. 3.

Protective efficacies of the ΔglnA1 and ΔglnA1EA2 mutants compared with BCG. C57BL/6 mice were immunized subcutaneously and challenged 90 days after immunization with virulent M. tuberculosis Erdman. The CFU numbers reflect the bacterial burden at 28 days post-aerosol challenge in the lungs and spleens in the first immunization experiment with the ΔglnA1 and ΔglnA1EA2 mutants (A) and the second experiment with one or two doses of the ΔglnA1 mutant (B). (C) Percent survival is shown for C57BL/6 mice (n = 8) immunized once or twice with the ΔglnA1 mutant, mice immunized once with BCG-Pasteur, and naive mice. (D) Histopathology of the lungs of vaccinated and unvaccinated mice 4 weeks after an aerosol challenge with M. tuberculosis Erdman. The error bars represent standard errors. *, P < 0.05; **, P < 0.001; determined by one-way ANOVA.

In the second vaccination experiment, two groups of mice were vaccinated 4 weeks apart with one or two doses of the ΔglnA1 mutant (Fig. 3B). At 4 weeks postchallenge, bacterial numbers in the lungs and spleens of BCG-vaccinated mice were similar to those of ΔglnA1-vaccinated mice, which is significantly lower than those in the unvaccinated controls (P < 0.05). Thirty days after the aerogenic challenge with virulent M. tuberculosis, differences were also apparent upon histopathological examination of the lungs. More severe spreading lung lesions in unvaccinated mice contrasted with those in vaccinated mice. Affected areas of the lung showed granulomatous pneumonia and were consolidated, with loss of air spaces (Fig. 3D). The lungs of mice vaccinated with the ΔglnA1 mutant were histologically similar to those of the BCG-vaccinated animals, with greatly reduced lung lesions.

By day 340 postinfection, all unvaccinated mice had succumbed to disease (MST, 297 ± 25 days). In contrast, and consistent with published results, BCG vaccination slowed the growth of wild-type M. tuberculosis in all organs examined and prolonged the survival of the mice by several months. Likewise, mice immunized with two doses of the ΔglnA1 mutant also exhibited enhanced survival (MST, 434 ± 25) that was equivalent to that of the BCG-vaccinated group (MST, 422 ± 19). The MST of animals immunized with one dose of the ΔglnA1 mutant was 344 ± 20 (Fig. 3C). The mean survival times for all vaccinated animals were statistically significant (P < 0.05).

In sum, we have demonstrated that three glutamine synthetases (GlnA2, GlnA3, and GlnA4) are not essential for the growth of M. tuberculosis in mice under in vitro and in vivo conditions. However, we have shown that the glnA1 gene is essential for growth of tuberculosis bacilli in the mouse model. Moreover, we have found that ΔglnA1 and ΔglnA1EA2 mutants are glutamine auxotrophs and that these mutants are potential live-vaccine candidates. Both mutants appear to be safe for vaccines because they do not replicate in mice and yet protect mice as effectively as BCG against an aerogenic M. tuberculosis challenge. It should be noted that live, nonreplicating attenuated strains of M. tuberculosis have generally been shown to elicit a superior long-term memory immune response, an advantageous vaccine characteristic (15, 22, 24). To enhance the safety of these live attenuated strains, public health officials and regulatory authorities have recommended that these vaccines be created with multiple unlinked deletions (10). A live-vaccine strain with at least two gene disruptions should be safer, since it is very unlikely that either a second-site suppressor mutation or a recombination event leading to loss of attenuation at two separate loci could occur. Recently, the ΔlysA ΔpanCD and ΔleuD ΔpanCD double auxotrophic mutants have been shown to be potential human vaccine candidates (21, 24), as the mutants are highly attenuated and immunogenic in vivo. In a similar manner, if additional appropriate unlinked gene deletions are made in the glutamine auxotrophic mutants, ΔglnA1 and ΔglnA1EA2, then the resulting strains should be inherently safe and thus could be seriously considered as viable vaccine candidates for prophylaxis against tuberculosis.