Protecting against antimicrobial effectors in the phagosome allows SodCII to contribute to virulence in Salmonella enterica serovar Typhimurium

Abstract

Salmonella enterica serovar Typhimurium replicates in macrophages, where it is subjected to antimicrobial substances, including superoxide, antimicrobial peptides, and proteases. The bacterium produces two periplasmic superoxide dismutases, SodCI and SodCII. Although both are expressed during infection, only SodCI contributes to virulence in the mouse by combating phagocytic superoxide. The differential contribution to virulence is at least partially due to inherent differences in the SodCI and SodCII proteins that are independent of enzymatic activity. SodCII is protease sensitive, and like other periplasmic proteins, it is released by osmotic shock. In contrast, SodCI is protease resistant and is retained within the periplasm after osmotic shock, a phenomenon that we term "tethering." We hypothesize that in the macrophage, antimicrobial peptides transiently disrupt the outer membrane. SodCII is released and/or phagocytic proteases gain access to the periplasm, and SodCII is degraded. SodCI is tethered within the periplasm and is protease resistant, thereby remaining to combat superoxide. Here we test aspects of this model. SodCII was released by the antimicrobial peptide polymyxin B or a mouse macrophage antimicrobial peptide (CRAMP), while SodCI remained tethered within the periplasm. A Salmonella pmrA constitutive mutant no longer released SodCII in vitro. Moreover, in the constitutive pmrA background, SodCII could contribute to survival of Salmonella during infection. SodCII also provided a virulence benefit in mice genetically defective in production of CRAMP. Thus, consistent with our model, protecting the outer membrane against antimicrobial peptides allows SodCII to contribute to virulence in vivo. These data also suggest direct in vivo cooperative interactions between macrophage antimicrobial effectors.

FIG. 1.

Expression and steady-state levels of SodCI and SodCII in vivo. Strains contain a chromosomal lacZ fusion in which the reporter gene is inserted just downstream of the stop codon for either sodCI-6xhis or sodCII. (A) Bacteria were recovered from LB after 16 h or from macrophages (MΦ) after 16 h of infection. (B) Bacteria were recovered from LB after 16 h or from mouse spleens (M) after 4 days of infection. (C) Same as in panel A, except that the background strain was a ΔhtrA mutant. Whole-cell extracts were prepared from equal numbers of bacteria, separated by SDS-PAGE, and subjected to immunoblot analysis. Each set of panels represents a single gel for which the resulting nitrocellulose membrane was cut into sections and processed with primary antibody directed against the indicated protein. These results are representative of experiments performed 2 to 6 times. The strains used were JS905, JS906, and JS907.

FIG. 2.

Transcription of sodCII is regulated by RpoS but is not directly affected by zinc level. (A) β-Galactosidase activity in strains containing sodCII-lacZ or katE-lacZ fusion in wild-type (WT) or rpoS null background, grown overnight in LB. The strains used were JS531, JS908, JS909, and JS910. (B) β-Galactosidase activity in strains containing lacZ fusions to the indicated genes in znuA⁺ or znuA null background, grown overnight in LB containing either EDTA or the zinc chelator TPEN, as indicated. The strains used were JS911, JS531, JS912, JS909, and JS913. (C) β-Galactosidase activity in strains containing znuA-lacZ or sodCII-lacZ fusion in wild-type or zur null background, grown overnight in LB. The strains used were JS911, JS914, JS531, and JS915.

FIG. 3.

SodCI and SodCII activities in the presence of zinc chelator. (A) Strains producing SodCI or SodCII from a plasmid (in an otherwise sod null background) were grown overnight in LB with the indicated concentration of the zinc chelator TPEN. SOD-specific activity was determined from whole-cell extracts. (B) Wild-type and znuA null strains were grown overnight in LB with the indicated concentration of TPEN, and the resulting OD₆₀₀ was determined. The strains used were JS916, JS917, 14028, and JS918.

FIG. 4.

Polymyxin causes release of periplasmic proteins. The left panel shows the results of osmotic shock (O) as described in Materials and Methods. The right panel shows the results of treatment at the indicated concentration of polymyxin B (PB). In each case, a whole-cell extract (W) was compared to the supernatant from an equivalent number of cells after treatment. Proteins in each sample were separated by SDS-PAGE and subjected to immunoblot analysis with primary antibody directed against the indicated protein. The strains used were JS921, JS922, and JS923.

FIG. 5.

SodCII is preferentially released by polymyxin B and CRAMP. Strains producing SodCI, SodCII, or monomeric SodCI (mSodCI) from plasmids (in an otherwise sod-negative background) were treated with various concentrations of polymyxin B (A) or CRAMP (B). The SOD activity released into the supernatant was compared to that in a corresponding whole-cell extract. The strains used were JS924, JS922, and JS925.

FIG. 6.

SodCII is not release by polymyxin B or CRAMP in a constitutive PmrA background. (A) Wild-type (left) and pmrA(Con) (right) strains were treated with the indicated concentration of polymyxin B (PB) or CRAMP. In each case, a whole-cell extract (W) was compared to the supernatant from an equivalent number of cells after treatment. Proteins in each sample were separated by SDS-PAGE and subjected to immunoblot analysis with anti-SodCII antibody. The strains used were JS926 and JS927. (B) β-Galactosidase activity in strains containing pmrI-lacZ or sodCII-lacZ fusion in wild-type or pmrA(Con) background, grown overnight in LB. PmrA does not control the transcription of sodCII. The strains used were JS928, JS929, JS906, and JS930.