The Proteasome of Mycobacterium tuberculosis Is Required for Resistance to Nitric Oxide

See allHide authors and affiliations

Science  12 Dec 2003:
Vol. 302, Issue 5652, pp. 1963-1966
DOI: 10.1126/science.1091176


The production of nitric oxide and other reactive nitrogen intermediates (RNI) by macrophages helps to control infection by Mycobacterium tuberculosis (Mtb). However, the protection is imperfect and infection persists. To identify genes that Mtb requires to resist RNI, we screened 10,100 Mtb transposon mutants for hypersusceptibility to acidified nitrite. We found 12 mutants with insertions in seven genes representing six pathways, including the repair of DNA (uvrB) and the synthesis of a flavin cofactor (fbiC). Five mutants had insertions in proteasome-associated genes. An Mtb mutant deficient in a presumptive proteasomal adenosine triphosphatase was attenuated in mice, and exposure to proteasomal protease inhibitors markedly sensitized wild-type Mtb to RNI. Thus, the mycobacterial proteasome serves as a defense against oxidative or nitrosative stress.

Mtb persistently infects about two billion people. The identification of pathways used by the microbe to resist elimination by the host immune response may suggest new targets for prevention or treatment of tuberculosis. During latent infection, the primary residence of Mtb is the macrophage. The antimicrobial arsenal of the activated macrophage includes inducible nitric oxide synthase (iNOS or NOS2) (1). At the acidic pH (≤5.5) prevalent in the phagosome of activated macrophages (2), nitrite, a major oxidation product of NO, is partially protonated to nitrous acid, which dismutates to form NO and another radical, ·NO2 (3). Thus, mildly acidified nitrite is a physiologic antimicrobial system. RNI may inflict not only nitrosative but also oxidative injury, such as when NO combines with superoxide from bacterial metabolism to generate peroxynitrite (4). Reagent NO kills Mtb with a molar potency exceeding that of most antituberculosis drugs (5, 6). In humans and mice with tuberculosis, macrophages in infected tissues and airways express enzymatically active iNOS (79), and mice lacking iNOS cannot control Mtb infection (10). Despite the protective effects of RNI, a small number of viable mycobacteria usually persist for the lifetime of the infected host (11) and sometimes resume growth.

To identify Mtb genes required for resistance against RNI, we screened 10,100 transposon mutants individually for increased sensitivity to nitrite at pH 5.5 [supporting online material (SOM text)]. Twelve mutants were hypersensitive. To quantify their phenotype, bacteria were exposed to pH 5.5 with or without 3 mM NaNO2 for 6 to 7 days and then monitored in two ways. First, to assess both growth inhibition and killing (Fig. 1A), cultures were diluted 1: 4 in unacidified medium (pH 6.6). The final pH (6.5) decreased the generation of NO from residual nitrite. At this point, the optical density (OD) of the cultures was barely detectable. Increased OD resulting from bacterial growth was recorded 2 to 3 weeks later. In control cultures with no nitrite, the 12 mutants grew as well as the wild-type strain after exposure to pH 5.5. In contrast, after exposure to nitrite, the mutants barely grew at all (Fig. 1A). Second, to assess killing, the bacteria were plated on agar, and colonyforming units (CFU) were counted after 2 to 3 weeks. The degree to which individual mutants survived nitrite ranged from almost none to the same as the wild-type strain (Fig. 1A).

Fig. 1.

(A) Survival of wild-type Mtb H37Rv and 12 mutants after 6 days of exposure to pH 5.5 with or without 3mM NaNO2. (Upper panel) Treated cultures were diluted 1: 4 in nitrite-free medium near neutrality (pH 6.6), and OD was monitored 21 days later. OD immediately after the 1: 4 dilution was barely detectable. (Lower panel) Alternatively, treated bacteria were plated on agar and CFU were counted 14 to 20 days later. Open bars show CFU at the beginning of the 6-day treatment. Error bars show means + SD of triplicates. Genes are designated below the columns. Numbers in parentheses indicate the codon in which the transposon was inserted. (B) Alignment of genomic regions encoding proteasome components in Mtb (Mt) and two other Actinomycetes, Streptomyces coelicolor (Sc) and Rhodococcus erythropolis (Re). The figure is modeled on figure 4 in (20). Proteasome-associated genes are represented in black. Homologs are joined by dashed lines. Identity percentages are circled. Genes identified in our screen are shown as outlined arrows.

Of the 12 mutants, 5 had insertions in two genes encoding putative components of the proteasome. Proteasomes, which are essential in eukaryotes (12), are responsible for degrading proteins that have been irreversibly oxidized (13), such as by peroxynitrite (14). The eukaryotic 26S proteasome is composed of a 20S (700-kD) core capped by one or two 19S regulatory structures. The base of the cap is a ring of six adenosine triphosphatases (ATPases) of the AAA (ATPase associated with various activities) class that participate in unfolding substrates, and the lid of the cap contains proteins required for substrate recognition, binding, and deubiquitinylation (12). Much less is known about proteasomes in prokaryotes, where caps have not been identified and ubiquitin is exceptional (15). The only eubacteria known to contain proteasomes are Actinomycetes. Mtb, an Actinomycete, is predicted to encode one type of α subunit (from prcA) and one type of β subunit (from prcB) (Fig. 1B). Recombinant PrcA and PrcB from Mtb bound each other in a two-hybrid assay (Fig. 2A), as expected for components of the proteasomal core. However, proteasomes have no known functions in bacteria, in which intracellular proteolysis is carried out by four other ATP-dependent, compartmentalized proteases of the ClpAP/XP, HslUV, FtsH, and Lon families. Mtb is unusual for a bacterium because it lacks HslUV and Lon (16).

Fig. 2.

Characterization of proteasome-associated genes and their mutants. (A) Self-association of Rv2115c and mutual association of PrcB and PrcA. Escherichia coli lacking endogenous adenylate cyclase was transformed with two plasmids encoding either of two domains of Bordetella pertussis cyclase, T18 or T25, in fusion with Rv2115c (R), PrcA (A), PrcB (B), or no other protein [empty vector (E)]. Protein-protein interactions that reconstituted functional cyclase allowed cAMP-dependent expression of β-galactosidase (19). Error bars show means + SD for two experiments, each in duplicate. (B) Growth rates of the proteasome-associated mutants (dashed lines) and wild-type Mtb (solid line) in 7H9-ADNaCl medium. Means ± SD for three cultures of each strain. (C) Single-copy complementation of an Rv2115c mutant with a transposon insertion at codon 77 (Rv2115c-77). Mutant and wild-type strains were transformed with vector alone (pMV306) or with the wild-type allele of Rv2115c (pMV-mpa). Bacteria were exposed to 3mM nitrite with pH 5.5 for 7 days and then plated on agar to determine CFU. Error bars show means + SD of triplicates. (D to F) Susceptibility to (D) heat (45°C, 24 hours), (E) isoniazid (0.06 μg/ml, 24 hours), and (F) H2O2 (5 mM, 4 hours) compared among wild-type Mtb H37Rv and Rv2115c (with or without complementation), Rv2097c, and uvrB mutants. Surviving bacteria were plated to determine CFU. Bacteria in panel b of (D) to (F) were plasmid-transformed. [(D) to (F)] Error bars show means + SD (six replicates, two experiments).

Three of the Mtb proteasome-associated mutants had insertions in Rv2115c, a gene predicted to encode AAA ATPase–forming ring-shaped complexes (ARCs) homologous to those found in proteasome caps in eukaryotes (17). Homologs of Rv2115c are known only in organisms with 20S proteasomes (Fig. 1B). The Rv2115c homolog in Rhodococcus erythropolis (82% identity) forms hexameric or heptameric rings compatible with capping a 20S proteasome core (17). Another homologous ATPase ring structure, proteasome activating nucleotidase (PAN) from the archaeon Methanococcus janaschii, stimulated ATP-dependent unfolding and proteasomal degradation of proteins in vitro (18). Rv2115c conserves Walker A and B boxes and other motifs characteristic of proteasome-associated ATPases (17, 18). Twohybrid analysis (19) demonstrated that Rv2115c monomers interact with each other, as expected for an ARC (Fig. 2A) (SOM text). Two mutants contained insertions in Rv2097c, a gene that is also annotated as a putative component of the proteasome (16, 20) in part on the basis of its association with prcBA core genes (Fig. 1B) (SOM text). All three Rv2115c mutants and both Rv2097c mutants grew at the same maximal rates and to the same final culture density as wild-type Mtb under standard culture conditions (Fig. 2B) (SOM text).

Complementation with an integrative plasmid encoding wild-type Rv2115c restored to the Rv2115c-77 mutant the wild-type resistance to acidified nitrite (Fig. 2C) (SOM text). Notwithstanding its sensitivity to acidified nitrite, the Rv2115c mutant was just as susceptible as the wild-type and complemented strains to heat and isoniazid (Fig. 2, D and E). However, the Rv2115c mutant was more resistant to H2O2 than either wild-type Mtb or the complemented mutant (Fig. 2F), perhaps reflecting compensatory induction of other antioxidant pathways. If Rv2097c participates in the same pathway as Rv2115c, it is likely that an Rv2097c mutant would have a similar signature of H2O2 resistance as the Rv2115c mutant. This was indeed the case (Fig. 2F). In contrast, a uvrB mutant was no more or less susceptible to H2O2 than wild-type Mtb (Fig. 2F).

Unlike wild-type Mtb, Rv2115c and Rv2097c mutants failed to grow in resting primary macrophages from wild-type (Fig. 3A) or iNOS–/– mice (Fig. 3B). Thus, the disruption of Rv2115c and Rv2097c sensitized Mtb to more macrophage-associated stresses than those dependent on iNOS.

Fig. 3.

Failure of Rv2115c and Rv2097c mutant Mtb to grow in wild-type and iNOS–/– macrophages; decreased growth and pathology of Rv2115c mutant Mtb in mice. (A and B) Bone marrow-derived macrophages from C57BL/6 wild-type mice (A) and iNOS–/– mice (B) were infected with Mtb H37Rv, Rv2115c-77, or Rv2097c-282. At the times shown, monolayers were washed, macrophages were lysed, and bacteria were plated for CFU. Error bars show means of triplicates + SD from one of three similar experiments. (C) Lungs 56 days after aerosol infection. (D) Magnified × 4 view of sections of the same lungs stained with hematoxylin and eosin. (E to G) Recovery of viable Mtb from the lungs (E), spleens (F), and livers (G) of mice at day 1 (n = 3), day 21 (n = 4), and day 56 (n is between 3and 4) after aerosol infection and upon euthanasia of moribund iNOS–/– mice (asterisk) on day 61 (n = 5) (error bars show means + SD). [(E), (F), and (G), lower panels] Effect of complementing the Rv2115 mutant Mtb by inserting a wild-type copy of the Rv2115c gene into the chromosome at the AT T site. The same plasmid without Rv2115c was also inserted into wild-type Mtb and the Rv2115c mutant (SOM). WT, wild type.

Wild-type and iNOS–/– mice were infected by inhalation of wild-type Mtb or an Rv2115c mutant (SOM text). By 8 weeks, tuberculous nodules were prominent in the lungs of wild-type mice infected with wild-type Mtb, but the Rv2115c mutant was markedly attenuated (Fig. 3C). Mice lacking iNOS suffered nearly complete consolidation of the lungs when infected with wild-type Mtb. The Rv2115c mutant caused an intermediate degree of gross pathology in lungs of iNOS–/– mice (Fig. 3C). All iNOS–/– mice infected with wild-type Mtb succumbed by week 9, whereas the other three experimental groups remained alive at week 37 (fig. S2). Histologically, the Rv2115c mutant caused minimal pneumonitis in wild-type mice and more pneumonitis in iNOS–/– mice (Fig. 3D and fig. S3). However, only wild-type Mtb caused necrosis, and only in iNOS–/– mice (Fig. 3D). By 8 to 9 weeks, we recovered 1 to 4 log10 fewer CFU from the lungs, spleens, and livers of mice infected with the Rv2115c mutant than we recovered from mice infected with the wild-type strain of Mtb. Similar results were observed in mice infected with either the Rv2097c or Rv2115c-608 mutants (fig. S4). The deficit in viable bacteria in each organ was reversed by complementing the Mtb mutant with a single copy of the wild-type allele of Rv2115c (Fig. 3, E to G). Thus, despite its normal growth in log and stationary phases in culture, Mtb lacking Rv2115c was far less fit than wild-type Mtb to grow in mice, to elicit pathology, and to kill the host. The decreased ability of the Rv2115c mutant to cause pathologic changes in the lungs was compensated in part by genetic inactivation of mouse iNOS. In sum, Mtb needs Rv2115c in order to withstand iNOS as well as other stresses imposed by the mammalian host.

We did not find prcBA mutants in our screen, probably because the prcBA operon is necessary for optimal growth or essential in Mtb (21), in contrast to its apparent dispensability in Mycobacterium smegmatis (22). As an alternative approach to study the role of prcBA in Mtb, we asked if chemical inhibitors of the chymotrypsin-like activity of the eukaryotic proteasome would reproduce the RNI-sensitive phenotype of the Rv2115c and Rv2097c mutants. Indeed, an inhibitor of the human proteasomal protease, N-[4-morpholine]carbonyl-β-[1-naphthyl]-L-alanine-L-leucine boronic acid (MLN-273) (12), potently blocked proteasomal protease activity in Mtb lysates. Its enantiomer, MLN-293, was inactive (Fig. 4A). Moreover, MLN-273 [but not MLN-293 or the vehicle, dimethyl sulfoxide (DMSO)] suppressed the growth of Mtb under standard culture conditions (Fig. 4, B to D). Lysates of the Rv2115c and Rv2097c mutants had normal proteasomal protease activity against the tetrapeptide substrate (fig. S5) and MLN-273 suppressed the growth of these mutants to same extent as for wild-type Mtb (Fig. 4, B to D). Finally, MLN-273 enhanced the ability of nitrite to suppress outgrowth of Mtb in culture (fig. S6). In survival assays based on the growth of Mtb on agar plates, an irreversible proteasomal protease inhibitor, epoxomicin (12), augmented the mycobactericidal effect of nitrite, but MLN-273, a reversible inhibitor, did not (Fig. 4E). In contrast, both MLN-273 and epoxomicin increased the antimycobacterial activity of nitrite when Mtb was given time to recover during a 4-day period of subculture at pH 6.5 in the presence of the inhibitors before being plated (Fig. 4F). In fact, both compounds were just as effective if added only at the time of the subculture (Fig. 4G). MLN-273 also augmented the toxicity of nitrite against the Rv2115c mutant (Fig. 4H). The control compound, MLN-293, had little or no activity under any conditions. Because the proteasomal protease inhibitors acted subsequent to nitrite-mediated injury, they appeared to block the ability of Mtb to recover.

Fig. 4.

Reproduction by proteasome inhibitors of the RNI-sensitive phenotype of the Rv2115c and Rv2097c muants. (A) MLN-273 but not its enantiomer MLN-293 inhibited proteasomal protease activity in Mtb lysates against a tetrapeptide substrate. (B to D) MLN-273 but neither MLN-293 nor the vehicle (DMSO) inhibited growth of wild-type Mtb and Rv2115c and Rv2097c mutants under standard culture conditions. MLN-273, MLN-293 (50 μM each), or DMSO (2% v/v) were added on day 0. (E) Epoxomicin (epoxo, 50 μM), an irreversible proteasome inhibitor, but not MLN-273 (273) (100 μM), a reversible inhibitor, augmented the antimycobacterial effect of nitrite when the inhibitors and nitrite were removed simultaneously by plating bacteria on agar after 6 days of exposure. (F and G) Both proteasome inhibitors augmented the antimycobacterial effect of nitrite if present after nitrite-mediated injury. Epoxomicin and MLN-273 but not MLN-293 enhanced the antimycobacterial effect, both when added along with nitrite at day 0 (F) and when added only after the subculture on day 6, plating on day 10 (G). (H) MLN-273 [25 μM, added on day 6 as in (G)] augmented nitrite-mediated injury of both wild-type and Rv2115c mutant Mtb. [(E) to (H)] Errors bars show means + SD for triplicates in one of two similar experiments.

Thus, two chemically distinct proteasome inhibitors produced a phenotype that both mimicked and augmented the effects of mutations in Rv2115c and Rv2097c. This supported the inferences from bioinformatics that Rv2115c and Rv2097c participate in proteasome function and do so at a separate site from the proteasomal protease. Although proteasome regulatory cap structures have not been identified in microbes, Rv2115c and Rv2097c may contribute to an analogous function. On the basis of its genomic organization and homology to ARC and PAN, we propose that Rv2115c be named mpa, for mycobacterial proteasome ATPase. Rv2115c may help unfold proteins and translocate them into the proteolytic core. We propose that Rv2097c be named paf, for proteasome accessory factor. Rv2097c may recognize a signal on a protein targeted for degradation.

This work shows that at least six different pathways are individually essential and nonredundant for resistance of Mtb to acidified nitrite (SOM text). Among these is the proteasome. Thus, one function of the bacterial proteasome is to protect the organism against oxidative or nitrosative stress. The mechanism of protection probably involves the degradation of proteins that are irreversibly oxidized, nitrated, or nitrosated (SOM text). Inhibition of the Mtb proteasome markedly sensitized the pathogen to bactericidal chemistries of the host. Specific inhibitors of the bacterial proteasome might be useful to sensitize Mtb to the immune system, especially if they are combined with agents that target one or more of the other RNI-resistance enzymes identified in this screen, such as UvrB.

Supporting Online Material

Materials and Methods

SOM Text

Figs. S1 to S6


References and Notes

View Abstract

Stay Connected to Science

Navigate This Article