Outer membrane permeability of the antibiotic-supersusceptible lipid A mutants of Escherichia coli to hydrophobic steroid probes

J Antimicrob Chemother 1999; 43: 608–610 Sir, Many clinically important Gram-negative bacteria possess effective outer membrane (OM) permeability barriers that markedly limit the penetration and, thus, the activities of hydrophobicantibiotics, such as erythromycin and rifampicin, and large antibiotic molecules, such asvancomycin. 1,2 This, in part, explains why these pathogens are resistant to most of the recently discoveredagents (e.g. the oxazolidinones and everninomicin) that exhibit potent activities against Gram-positive bacteria. 3 The barrier function of the OM relies principally on the compact leaflet of lipopolysaccharide (LPS) molecules that covers the cell surface and which is fairly impermeable to hydrophilic and (moderately) hydrophobic compounds. This is exemplified by the recently described Escherichia coli mutants, lpxA and lpxD, in which the lipid A moieties of the LPS have undergone profound alterations and which arealmost as susceptible to hydrophobic antibiotics as the Gram-positive reference strains. 3,4 The MICs of a large number of antibiotics for the E. colilpxA mutant are 30- to 1000-fold lower than those for clinical isolates of E. coli. 4 Other LPS mutants, such as those in which the inner core polysaccharide moieties of the LPS aredefective (Re type), also exhibit antibiotic supersusceptibility, although to a lesser extent than thelipid A mutants. The most reliable method of quantifying the permeability of the OM to hydrophobic compounds is that developed by Plésiat & Nikaido 5 which allows the rates of penetration of various steroid probes across the lipid portion of the bilayer to be accurately determined. The results are expressed in terms of the permeability coefficient (P) in nanometres per second (nm/s). With this approach, these investigators showed that the OM of a Re mutant of Salmonella typhimurium was up to 25- and 16-fold more permeable to uncharged steroids and to testosterone hemisuccinate (a monoanionic probe) respectively than that of the parent strain. 5 P values for the most antibiotic-hypersusceptible strains, i.e. the lipid A mutants, have not yetbeen reported. In the present study, we demonstrate that the diffusion rate of testosteronehemisuccinate through the lpxA-type OM is much higher than that through the Re-type OM. The strains used in the study were as follows: the lipid A mutants of E. coli SM101 (lpxA) and E. coli CDH23-213 (lpxD, the gene formerly known as omsAfirA and ssc), both of which are thermosensitive, i.e. SM101 grows well at 28°C, but not at 37°C, andCDH23-213 grows well at 20°C and 37°C, but not at 42°C; the corresponding isogenic parent-type strains, SM105 (lpxA+) and CDH23-210 (lpxD+), which were used as controls; and the E. coli mutant strain, D21f2 (rfa), which has a defective inner core oligosaccharide and which elaborates Re-type (heptoseless)LPS. The antibiotic susceptibilities of all of the strains have been characterized previously. 3,4 The study strains were transformed with the recombinant plasmid, pLE689, that mediates the 3-oxosteroid Δ1-dehydrogenase enzyme of Comamonas testosteroni. 5 The media used were LB agar and broth, both containing 10 g of tryptone (Oxoid, Basingstoke,UK), 5 g of yeast extract (Difco, Detroit, MI, USA) and 5 g of NaCl, at pH 7.2, and supplemented with kanamycin at a concentration of 25 mg/L. The agar plates and broth cultures were incubated at 30°C. The steroid permeability assay was performed as described previously. 6 Briefly, cells in the exponential growth phase were collected by centrifugation at roomtemperature and resuspended in 50 mM HEPES, pH 7.4, to an A650 of 3. Tubes containing 700 µL of bacterial suspension diluted three-, six- or 30-fold in the same buffer were incubated at30°C with gentle agitation. The neutral steroid, testosterone, or its hemisuccinate derivative was added at time 0 to give final concentrations of 50 µM and 200 µM respectively. Following incubation for 10 min, >95% of the steroid was removed by two brief extractions with ethyl acetate. The concentration ofthe Δ-dehydrogenated product was then determined by high-performance liquid chromatography (HPLC) and the permeability coefficients were calculated according to an equation based on Ficks' First Law of Diffusion. 6 The P values for testosterone hemisuccinate in the lipid A mutants, SM101 (lpxA) and CDH23-213 (lpxD), were approximately 900–1100 nm/s—approximately five- to six-fold greater than that in the E. coli Re mutant, D21f2 (Table), approximately ten-fold greater than that previously determined in the Re mutant of S. typhimurium (data not shown) and approximately 50- to 100-fold greater than those in the control strains (Table). Of considerable note, the high P values for the lipid A mutants demonstrated here are comparable to that (approximately 1000 nm/s) previously determined for the wild-type strain, S. typhimurium SL696, 5 following exposure to deacylpolymyxin B (DAPB), a potent permeabilizer of the OM. 7 This suggests that the OM permeability barrier to a representative anionic hydrophobic probemolecule is quantitatively equally defective in both the lipid A mutants and theDAPB-permeabilized cells. The P values for testosterone in the lipid A mutants (3260 nm/s and 4270 nm/s inCDH23-213 and SM101, respectively) were similar to that (3620 nm/s) in the Re mutant (Table). This suggests that the Re-type OM is already maximally permeable to this neutral hydrophobic probe. In accord with this observation, previous investigators have shown that the P coefficient for another uncharged steroid, androstanedione, in DAPB-treated S. typhimurium was comparable to that in the Re mutant of S. typhimurium.5 Studies such as the one described here add to our knowledge of the properties of the permeability barriers of mutationally defective OMs. This in turn provides information that might help in designing novelantibiotics with activities against Gram-negative bacteria. In the light of increasing antibioticresistance worldwide, such agents are needed desperately. P values (nm/s) for testosterone and testosterone hemisuccinate in the lipid A mutants of E. coli,lpxA and lpxD, their isogenic parent-type strains (controls) and a mutant strain of E. coli, D21f2, with a defective inner core oligosaccharide P values (nm/s) for testosterone and testosterone hemisuccinate in the lipid A mutants of E. coli,lpxA and lpxD, their isogenic parent-type strains (controls) and a mutant strain of E. coli, D21f2, with a defective inner core oligosaccharide The excellent technical assistance of Colette Godard is gratefullyacknowledged.