Bicarbonate Resensitization of Methicillin-Resistant Staphylococcus aureus to β-Lactam Antibiotics

Abstract

Endovascular infections caused by methicillin-resistant Staphylococcus aureus (MRSA) are a major health care concern, especially infective endocarditis (IE). Standard antimicrobial susceptibility testing (AST) defines most MRSA strains as "resistant" to β-lactams, often leading to the use of costly and/or toxic treatment regimens. In this investigation, five prototype MRSA strains, representing the range of genotypes in current clinical circulation, were studied. We identified two distinct MRSA phenotypes upon AST using standard media, with or without sodium bicarbonate (NaHCO₃) supplementation: one highly susceptible to the antistaphylococcal β-lactams oxacillin and cefazolin (NaHCO₃ responsive) and one resistant to such agents (NaHCO₃ nonresponsive). These phenotypes accurately predicted clearance profiles of MRSA from target tissues in experimental MRSA IE treated with each β-lactam. Mechanistically, NaHCO₃ reduced the expression of two key genes involved in the MRSA phenotype, mecA and sarA, leading to decreased production of penicillin-binding protein 2a (that mediates methicillin resistance), in NaHCO₃-responsive (but not in NaHCO₃-nonresponsive) strains. Moreover, both cefazolin and oxacillin synergistically killed NaHCO₃-responsive strains in the presence of the host defense antimicrobial peptide (LL-37) in NaHCO₃-supplemented media. These findings suggest that AST of MRSA strains in NaHCO₃-containing media may potentially identify infections caused by NaHCO₃-responsive strains that are appropriate for β-lactam therapy.

Keywords: MRSA; antimicrobial susceptibility testing; infective endocarditis; penicillin-binding proteins; sodium bicarbonate; β-lactams.

FIG 1

Time-kill analysis of log-phase cells grown in media with or without NaHCO₃ . (A) MRSA 11/11; (B) COL. Growth in CA-MHB-Tris (closed symbols) or in CA-MHB-Tris plus 44 mM NaHCO₃ (open symbols) with no drug (circles), 8 μg/ml cefazolin (triangles), 32 μg/ml cefazolin (diamonds), or 15 μg/ml oxacillin (squares) is indicated. The data are means of two independent runs performed in triplicate for each condition ± the SD. Statistical comparisons were made using a Kruskal-Wallis single-factor ANOVA and the post hoc pairwise Mann-Whitney U test. Asterisks represent comparisons (**, P < 0.01) at the 24-h time point of MRSA 11/11 exposed to 8 μg/ml cefazolin in CA-MHB-Tris versus CA-MHB-Tris plus 44 mM NaHCO₃ and 15 μg/ml oxacillin in CA-MHB-Tris plus 2% NaCl versus CA-MHB-Tris plus 2% NaCl plus 44 mM NaHCO₃.

FIG 2

Population analysis of cells grown in media with or without NaHCO₃. (A) Mueller-Hinton agar (MHA) supplemented with Tris (closed symbols); (B) MHA supplemented with Tris and 44 mM NaHCO₃ (open symbols). Data for strains MRSA 11/11 (solid black line), MW2 (dashed black line), 300-111 (dotted gray line), COL (solid gray line), and BMC1001 (dashed gray line) are shown. CFZ, cefazolin. Area under the curve (AUC) ratios were compared to COL as a homoresistant reference strain. The data are means of two independent runs performed in triplicate for each condition ± the SD. Statistical comparisons were made using a Kruskal-Wallis single-factor ANOVA and the post hoc pairwise Mann-Whitney U test. MRSA 11/11, MW2, and 300-111 AUC in media containing 44 mM NaHCO₃ were significantly reduced compared to strains in media without NaHCO₃ (*, P < 0.05; **, P < 0.01; ****, P < 0.0001); the AUC ratio of 300-111 in media containing 44 mM NaHCO₃ is significantly greater than that of MRSA 11/11 and MW2 (*, P < 0.05).

FIG 3

Treatment outcomes of rabbits with infective endocarditis treated with β-lactams. (A) NaHCO₃-responsive strains; (B) NaHCO₃-nonresponsive strains. All rabbits were treated with 100 mg/kg of cefazolin (CFZ) or oxacillin (OX), t.i.d., intramuscularly, for 4 days. Data for strains MRSA 11/11 (control, n = 6; CFZ, n = 8; OX, n = 6), MW2 (control, n = 7; CFZ, n = 6; OX, n = 7), COL (control, n = 6; CFZ, n = 7; OX, n = 6), BMC1001 (control, n = 6; CFZ, n = 7; OX, n = 7), and 300-111 (control, n = 10; CFZ, n = 4; OX, n = 6) are shown. The data presented indicate the mean tissue CFU/g for each treatment group ± the SD. Statistical comparisons were made using a Kruskal-Wallis single-factor ANOVA and the post hoc pairwise Mann-Whitney U test (**, P < 0.01). Strain 300-111 had significantly greater cell counts in all target tissues following cefazolin and oxacillin therapy compared to both MRSA 11/11 and MW2 (*, P < 0.05).

FIG 4

LL-37 synergy with β-lactams in media containing 25 mM NaHCO₃. (A) MRSA 11/11 (NaHCO₃ responder); (B) MW2 (NaHCO₃ responder); (C) COL (NaHCO₃ nonresponder). CFZ, cefazolin; OX, oxacillin. Bars: no drug, hatched; LL-37 only, white; CFZ or OX, black; CFZ or OX plus LL-37, gray. MRSA 11/11 was exposed to 2.5 μg/ml LL-37 and 0.03125 μg/ml CFZ or OX. MW2 and COL were exposed to 5 μg/ml LL-37 and 0.0625 μg/ml CFZ or OX. The percent survival was calculated after 4 h of exposure to antimicrobials in NaHCO₃-free DMEM supplemented with 25 mM NaHCO₃. The data are means of three independent runs performed in triplicate for each condition ± the SD. Statistical comparisons were made using a Kruskal-Wallis single-factor ANOVA and the post hoc pairwise Mann-Whitney U test (**, P < 0.01; ***, P < 0.001).

FIG 5

mecA and sarA expression in NaHCO₃-responsive and NaHCO₃-nonresponsive strains. (A) qRT-PCR analysis of mecA and sarA gene expression in stationary-phase cells grown in CA-MHB-Tris with or without NaHCO₃. (B) qRT-PCR analysis of mecA and sarA gene expression in stationary-phase cells grown in CA-MHB-Tris plus 2% NaCl with or without NaHCO₃ with 1/2 MIC of oxacillin. Gene expression levels are normalized to the housekeeping gene gyrB. Gene expression values are normalized to CA-MHB-Tris for each strain. All data are derived from two independent biological replicates, tested in triplicate on two separate occasions. (B) Lipase production analysis in NaHCO₃-responsive and -nonresponsive strains grown in media with or without NaHCO₃. Values normalized to CA-MHB-Tris for each strain. The data are means of two independent runs performed in triplicate for each condition ± the SD. Statistical comparisons were made using a Kruskal-Wallis single-factor ANOVA and the post hoc pairwise Mann-Whitney U test (*, P < 0.05; **, P < 0.01; ****, P < 0.0001).