|Home | About | Journals | Submit | Contact Us | Français|
The bactericidal activity of Streptococcus pneumoniae toward Staphylococcus aureus is mediated by hydrogen peroxide. Catalase eliminated this activity. Pneumococci grown anaerobically or genetically lacking pyruvate oxidase (SpxB) were not bactericidal, nor were nonpneumococcal streptococci. These results provide a possible mechanistic explanation for the interspecies interference observed in epidemiologic studies.
Nasal Staphylococcus aureus and nasopharyngeal Streptococcus pneumoniae colonization serve as a source for infection of other sites and for transmission between humans (12, 24). Epidemiological studies have suggested that nasopharyngeal carriage of S. pneumoniae may inhibit nasal carriage of S. aureus (4, 17). Hydrogen peroxide produced by S. pneumoniae has been reported to be toxic in vitro to other bacteria (11, 15) and to mammalian cells (2, 5, 9). H2O2 is produced by pyruvate oxidase (SpxB) as a by-product of aerobic metabolism (19). We sought to test the hypothesis that S. pneumoniae directly inhibits S. aureus in vitro by an H2O2-dependent mechanism.
(The results of this study were partially presented at the 45th Interscience Conference of Antimicrobial Agents and Chemotherapy, Washington, D.C., October 2005.)
Bacterial strains used in this study are described in detail in Table Table1.1. S. pneumoniae strains used were TIGR4 (21), Rx1 (16), and Pn-20. In addition, SpxB-negative variants of TIGR4 and Pn-20 and LytA-negative variants of Rx1 and Pn-20 were created using the Janus cassette (20). Strains of three other Streptococcus spp.—S. gordonii, group A streptococcus, and group C streptococcus—were also tested. Staphylococcal species used in the study were the following: S. aureus strains Newman (NCTC 8178) and ALR and coagulase-negative staphylococci (CNS) (Staphylococcus epidermidis, Staphylococcus xylosus, and Staphylococcus sciuri). Bovine liver catalase (1,000 U/ml; MP Biomedicals, Inc., Solon, OH) was added to growth medium when specified. Bacterial H2O2 production was measured colorimetrically in culture supernatants (8).
To quantify bactericidal activity, we cocultured bacterial strains in brain heart infusion (BHI) and evaluated their survival on selective media. Serial twofold dilutions of the staphylococcal strain were mixed with twofold serial dilutions of the streptococcal strain in a final volume of 100 μl (see legend to Fig. Fig.11 for more details). Cultures with and without catalase were incubated at 37°C in 5% CO2. At time zero and 6 h, approximately 2 μl of each culture was plated on selective medium using a replica plater (Sigma, Aldrich).
For a given S. aureus inoculum, a higher S. pneumoniae inoculum led to greater killing (Fig. 1A and B). The diagonal pattern of killing (Fig. (Fig.1B)1B) suggests that the inoculum of S. aureus that could be killed completely was approximately linearly proportional to the pneumococcal inoculum. Killing was eliminated by the addition of catalase (Fig. 1C to D). No inhibition of S. pneumoniae by S. aureus was observed (Fig. 1E to F).
To quantify bactericidal efficiency, we report the maximum staphylococcal inoculum “killed” (reduced below the detection limit of 50 CFU/well) by 106 CFU of streptococci at 6 h (IK6). Tables Tables22 and and33 compare the killing efficiencies of different streptococci and the susceptibilities of different staphylococci. SpxB-negative strains did not produce detectable amounts of H2O2 and were minimally bactericidal to S. aureus. The maximum inoculum of S. aureus killed by S. pneumoniae Pn-20ΔspxB was 1.2 log10 CFU (IK6 = 1.2), 5 orders of magnitude lower than that of the parental strains (Table (Table2);2); catalase abolished the residual killing activity. Other streptococcal species that did not produce detectable levels of H2O2 demonstrated little or no bactericidal effect (Table (Table22).
To verify that S. aureus is inhibited by a soluble factor produced by S. pneumoniae and that this factor is H2O2, we evaluated the growth of S. aureus (Newman) in filtered pneumococcal supernatant and compared it to growth in fresh BHI and in coculture with S. pneumoniae (Pn-20). Figure Figure22 demonstrates a bactericidal effect of sterile supernatant from aerobic cultures of S. pneumoniae to S. aureus within 6 h (Fig. (Fig.2A).2A). This effect was mitigated either when catalase was added to the medium or with supernatant from anaerobically grown S. pneumoniae, which does not produce H2O2 (14) (Fig. (Fig.2B2B).
S. pneumoniae variants lacking the major autolytic enzyme LytA exhibited bactericidal activities similar to those of their parent strains (Table (Table2).2). This finding excludes a potential accessory role for this toxic enzyme, while also serving as a control to demonstrate that the significantly reduced bactericidal activity of ΔspxB strains was a direct result of deficient SpxB activity and not by the genetic manipulation itself.
S. aureus Newman was more susceptible to S. pneumoniae's bactericidal effect than strain ALR. CNS strains showed undetectable inhibition by S. pneumoniae (Pn-20) in cocultures (IK6< 2). These results correlated (Table (Table3)3) with the susceptibilities of the staphylococcal strains to H2O2 added to the medium (1 μmol every 90 min) as well as to the catalase activities of these strains measured with the Amplex Red catalase assay kit (Molecular Probes, Eugene, OR). External H2O2 added periodically reduced CFU/ml of S. aureus strains by 1.5 to 2.5 log10, while the neutralization of this external H2O2 by exogenous catalase permitted an increase of 1.5 log10 CFU/ml. In contrast, under the same conditions (with no addition of catalase), the numbers of CFU/ml for CNS strains increased slightly (0.5 log10) (Table (Table33).
It is striking that S. aureus, a catalase-positive species, is so susceptible to a H2O2-producing bacteria. The two different S. aureus strains tested were highly susceptible to S. pneumoniae, while CNS species were highly resistant to S. pneumoniae. All staphylococci elaborate catalase, with variation between species in the number and expression of catalase enzymes (1, 3). Indeed the staphylococcal species that were resistant to S. pneumoniae secreted higher concentrations of catalase than S. aureus; however, the concentrations detected may not linearly reflect differences in intracellular catalase, and the greater antioxidant capacity of these species might be due to additional factors not measured in this study.
Our impetus for this study was the epidemiologic data showing an inverse association between S. aureus and S. pneumoniae colonization (4, 17). H2O2-mediated, unidirectional antagonism as observed here in vitro provides one possible mechanism for this inverse association. If the association is causal and acquisition of S. pneumoniae eradicates S. aureus carriage, then use of pneumococcal vaccines may eliminate the “protective” effect of S. pneumoniae against S. aureus carriage and an increase in S. aureus carriage will follow. Increased S. aureus otitis media has been observed among vaccinees in a pneumococcal conjugate vaccine randomized trial (23). Whether the current increase in severe community-acquired S. aureus infections, including methicillin-resistant S. aureus (6), is partially caused by the recent introduction of the pneumococcal conjugate vaccine is yet to be determined.
This study provides only the first step towards elucidating the mechanism of S. pneumoniae-S. aureus interference in vivo. In vivo interference could be caused by a direct bactericidal effect similar to that observed here. Alternatively, or in addition, it could occur via a host cell signaling pathway. Interestingly, such an indirect pathway might also involve H2O2, which plays key roles in cell signaling, particularly in regulation of the production of cytokines (7, 25). Future in vivo studies are required to address these important remaining questions.
We thank Michael Wessels for providing Streptococcus group A, Jean Lee for providing the staphylococcus strains used in the study and for critical review of the manuscript, and Amit Srivastava for thoughtful advice.
This study was supported by NIH grant 1R01AI48935 to M.L. and 1R01AI067737-01 and the Pamela and Jack Egan Fund to R.M.