Mycobacteria are characterized by long-term survival in the macrophage. Understanding this lifestyle is crucial to answering key questions about mycobacterial pathogenesis. The bacilli infect macrophages early in mycobacterial disease, where they remain protected from specific and nonspecific immune responses and many antibacterial drugs. Yet little is known about intracellular nutrition, i.e., the sources of carbon, nitrogen, and energy and how they are acquired.
Analysis of the phenotypes of peptide permease mutants is often difficult due to the presence of more than one permease with overlapping specificities. In addition, the lack of availability of radiolabeled peptides makes kinetic analysis difficult and costly. Toxic peptides are useful tools for the selection and characterization of peptide permeation mutants. Here, glutathione (γ-glutamyl-l-cyteinylglycine [GSH]) was shown to be toxic to BCG. GSH and its toxic NO derivative,S-nitrosoglutathione (GSNO), were used to characterize a peptide permease mutant of BCG.
DISCUSSION
A mutant strain of BCG was constructed in which the
oppD gene was interrupted with a selectable marker. Sequence homology indicates that
oppD encodes the ATPase component of a binding protein-dependent transport system. This component is positioned as a dimer of peripheral membrane proteins localized on the cytoplasmic side of the membrane. These proteins contribute no substrate specificity to the transport system; rather, they function to couple ATP hydrolysis to translocation of substrate across the membrane. Downstream of this ORF, separated by 4 bp, is the
oppA homolog, the substrate binding protein. The number of binding proteins in ABC transport systems usually exceeds that of the membrane components, at ratios of as high as 30- to 50-fold, as is the case with the maltose- and histidine-binding proteins (
4). The alterations in peptide uptake exhibited by the BCG
oppmutant suggest that the
oppD insertion exerts a polar effect on
oppA gene expression as well, although this remains to be confirmed by, for example, immunodetection.
Two approaches to phenotypic characterization of the BCG(
oppΔ) mutant were taken. Utilization of peptides as sole source of carbon or nitrogen can be compared between the wild-type and mutant strains.
Mycobacterium smegmatis can use a wide range of peptides to support growth (
2). But wild-type BCG was unable to grow on 25 peptides supplied as the sole carbon source, and extremely poor growth was observed on a small number of peptides supplied as the sole nitrogen source. The inability of BCG to grow on peptides as the sole carbon or nitrogen source may be due to insufficient uptake. Transport of several amino acids has been shown to be limiting for growth in
E. coli and
Klebsiellaspp., and mutations resulting in increased transport of the substrate in question can overcome the negative growth phenotype (
28). Whether the inability of BCG to utilize peptides for growth is due to insufficient transport can be tested by overexpression of the
opp system.
A fruitful approach to the characterization of nutrient transport mutants is to examine resistance to toxic substrate analogs. Two toxic peptides commonly used with bacteria are triornithine, to which mycobacteria are entirely insensitive (N. D. Connell, unpublished data), and bialaphos (
l-alanylalnylphosphothricin). Bialaphos is a tripeptide comprised of two alanine residues and a phosphothricin moiety. After transport by the
opp system, bialaphos is cleaved by intracellular peptidases. The phosphothricin moiety is a glutamate analog that binds irreversibly to glutamine synthetase and kills the cell.
opp mutants of
B. subtilis (
43) and
Streptomyces coelicolor(
34) are resistant to bialaphos. Mycobacteria are sensitive to bialaphos, and the
opp mutant described here is fully sensitive to the drug (A. Bhatt and N. D. Connell, unpublished data). A likely interpretation of this result is that the BCG Opp characterized in this study does not transport the tripeptide bialaphos.
The tripeptide GSH, found in most living cells, was used here to characterize the BCG(
oppΔ-19) mutant. A range of functions has been attributed to the peptide, including cofactor function, acting as a transporter component, providing an alternative source of sulfur, and participation in cellular processes such as DNA and protein synthesis, regulation of enzyme activity, and membrane function. In bacteria, GSH is found in facultative and aerobic bacteria but not in strict anaerobes (
37).
We have shown that, surprisingly, GSH is toxic to mycobacteria. Mycobacteria and other actinomycetes do not synthesize GSH. Rather, they produce mycothiol [1-
d-myo-inosityl-2-(
N-acetyl-
l-cysteinyl)amido-2-deoxy-alpha-
d-glucopyranoside; MSH] in millimolar amounts (
1). MSH has been isolated from a number of mycobacterial species, including
Mycobacterium smegmatis,
Mycobacterium bovis, and
M. tuberculosis H37Rv (
1,
5,
31,
45).
The basis of the toxicity of GSH to mycobacteria is unknown and not previously reported. One possibility is that the presence of high concentrations of GSH may result in an imbalance in a bacterium containing an alternative thiol for regulating reduction/oxidation activity (i.e., mycothiol).
Interestingly, GSH is similar in structure to penicillin precursors produced by
Penicillium and
Cephalosporium spp., and the β-lactam form of GSH is penicillin-like. Spallholz has hypothesized that GSH is an evolutionary precursor of antibiotics produced by higher eukaryotes before the emergence of cellular immunity (
44). Mycobacteria may possess some intrinsic sensitivity to this structure.
Nitric oxide (i.e., NO) and related reactive nitrogen intermediates are thought to be major antimicrobial agents produced during the host defense response (
23,
24,
29,
30). In view of the high levels (millimolar concentrations) of GSH found in mammalian cells, the nitrosothiol GSNO is a strong candidate for an in vivo NO donor. In a genetic screen for
Salmonella mutants resistant to GSNO, De Groote et al. recovered mutants with defects in
dppD and
dppA function (
10), homologs of the two genes interrupted in the BCG mutant described here.
We have shown that GSNO is cytocidal for wild-type BCG at concentrations similar to those to which Salmonella is sensitive (1 to 2 mM). While the data demonstrate reduced uptake of both GSH and GSNO by the opp BCG mutant, theS-nitrosothiol clearly kills both mutant and wild-type cells in culture. GSH, on the other hand, is toxic only to the wild-type strain. It is possible that GSNO, but not GSH, is transported by more than one permease. Alternatively, unlike Salmonella, BCG may be sensitive to extracellular NO provided by GSNO, and transport of the nitrosothiol, either as a tri- or dipeptide, is not required for toxicity.
Note that the two methods of analysis of toxicity used ([
3H] uracil incorporation, Tables
1 to
5, and growth, Fig.
5) yielded some inconsistencies in resistance levels. These are likely the result of differences between the two assays. [
3H]uracil incorporation is a general evaluation of the metabolic state of the cells, since [
3H]uracil is incorporated into metabolic pools, entering first RNA and later other macromolecules by degradation and reincorporation. The growth curves are a less sensitive indicator of the effects of these peptides on metabolism.
Uptake of GSNO in
Salmonella and
E. coli is dependent on the
ggt locus, encoding the γ-glutamyltranspeptidase (γGT) (
46). γGT transfers a γ-Glu group to an acceptor amino acid or peptide, releasing the dipeptide
S-nitrosocysteinylglycine. This points to the possibility that the NO dipeptide is the actual toxic species in the toxicity of GSNO against
Salmonella. γGT activity has been described in a number of species of mycobacteria (
M. smegmatis and
Mycobacterium avium) (
20,
40), and two ORFs homologous to the
ggt gene are present in the Sanger database (Rv0773c and Rv2394). The latter contains a clear lipoprotein motif, suggesting that a transpeptidase is localized at the cell surface in mycobacteria (
7).
The data in Tables
3,
4, and
5 suggest that the dipeptide
l-Cys-Gly may be at least partially responsible for the toxicity of GSH, since we have demonstrated a low-level toxicity of
l-Cys-Gly against BCG. The dipeptide is not active against the
opp mutant. Further, the data in Table
4 suggests that
l-Cys may be a toxic component of both the tri- and dipeptides. Since this amino acid enters cells by an as-yet-unidentified amino acid permease, all three strains are sensitive to its low-level toxicity. Finally, Gly has no effect on any of the three strains.
Several reports point to a crucial role of peptide transporters in microbial cell signaling and virulence. For example, di- and oligopeptide permeases were identified among a number of mutant loci affecting growth and survival of
Staphylococcus aureus in multiple infection environments (
9). In
Borreliaspp., peptide-binding proteins of ABC-type transporters were found to be conserved those species causing Lyme disease (
19). In the group A streptococci, the expression of the cysteine protease is reduced in a dipeptide permease mutant, and expression of the
dpp operon is under the control of the Mga virulence regulator (
38). The survival of the
opp mutant of BCG in cultured, unactivated murine macrophages was unimpaired compared with its wild-type parent (data not shown). This result suggests that, in this assay, access to small peptides is not a major nutritional requirement of BCG.
In addition to the
opp operon, the
M. tuberculosis database contains an operon (Rv3663c to Rv3666c) homologous to the
dpp operon of
E. coli and
B. subtilis. A tripeptide permease system (
tpp) is present in enteric bacteria (
3,
4) and
L. lacti (
11); a
tpp operon has not been identified in the
M. tuberculosis genome. The specificity of peptide transport systems has been established by transport studies combining single and multiple mutants, peptides, and their analogs (
27,
36). The Opp of enteric bacteria transports any peptide up to six amino acids (
13,
35), whereas the Opp of
L. lactis is restricted to four to eight amino acids (
21). The Dpp of enteric bacteria is specific for dipeptides, and the Tpp is specific for hydrophobic tripeptides (
15). The data presented in this study suggest that the Opp of BCG may resemble that of
E. coli and
Salmonella, transporting both di- and tripeptides.
An alternative interpretation is that the
opp designation assigned to this particular operon in the
M. tuberculosisH37Rv database is in error and that the operon described here, spanning ORFs Rv1280c to Rv1283c, actually encodes a dipeptide permease system. This possibility is suggested by four observations. First, mutations in
dpp confer resistance to GSNO in
Salmonella spp. (
10). Second, as mentioned earlier, two of the four genes in the
opp operon of
M. tuberculosis (Rv1283c and Rv1281c) show higher homology to
dpp components of other species than to
opp components (
7). Third, unlike
opp mutants of
Bacillus (
43) and
Streptomyces (
34), the BCG
opp mutant described here is not resistant to the toxic tripeptide bialaphos. Fourth, the BCG
opp mutant is resistant to the toxic effects of both GSH and
l-Cys-Gly.
Further studies are required to understand peptide discrimination of the mycobacterial permeases. To this end, our laboratory has cloned the second annotated peptide permease (dpp) (Rv3663c to Rv3666c) from BCG and is engaged in constructing two additional mutants (dppΔ and a double dppΔ oppΔ mutant) to complement studies with the oppΔ mutant described here. Finally, construction of such mutants in M. tuberculosis, currently under way, will enable the analysis of the role of peptide transport and metabolism in mycobacterial pathogenesis.