Rχ-01, a New Family of Oxazolidinones That Overcome Ribosome-Based Linezolid Resistance

Antibiotics.Compounds Rx-01_002, Rx-01_007, Rx-01_133, Rx-01_149, Rx-01_413, Rx-01_423, Rx-01_445, and Rx-01_667 (9, 34) and linezolid (7) (Fig. 1) were synthesized at Rib-X Pharmaceuticals, Inc., New Haven, CT. Chloramphenicol, gentamicin, and tylosin were obtained from Sigma (St. Louis, MO). [α-³³P]dTTP (3,000 Ci/mmol) was obtained from Perkin-Elmer, [³H]chloramphenicol (20 Ci/mmol) was obtained from American Radiolabeled Chemicals Inc. (St. Louis, MO), and [³H]puromycin (9.1 Ci/mmol) was obtained from Moravek (Brea, CA). Etamycin and griseoviridin, used as controls to validate the assay (data not shown), were a gift from G. S. Katrukha, Institute of New Antibiotics, Moscow, Russia.

• Open in new tab
• Download powerpoint

FIG. 1.

Chemical structures of Rχ-01 novel oxazolidinones and linezolid.

Bacterial strains. S. aureus ATCC 29213 and Enterococcus faecalis ATCC 29212 were obtained from the American Type Culture Collection. S. aureus RN1786, a nuclease-deficient strain and the source of wild-type ribosomes for translation, was obtained from S. Khan (University of Pittsburgh School of Medicine). S. aureus A7820 (Lin^r Erm^r) was derived from A7819, a linezolid-resistant (G2576U) strain, isolated in the clinic, to which the ermC methylase gene had been introduced (39) and was obtained from Robert Moellering, Jr. (Beth Israel Deaconess Medical Center, Boston, MA). E. coli MC245 strains containing plasmids with lacZ protein fusions engineered to test translational accuracy (48) were a generous gift from A. Dahlberg (Brown University, Providence, RI).

Preparation of S. aureus 70S ribosomes.Ribosomes were prepared according to the method of Rheinberger et al. (37) with small modifications to adapt the protocol for S. aureus. Briefly, an overnight culture of S. aureus was used to inoculate fresh tryptic soy broth medium. Bacteria were grown at 37°C to an optical density at 600 nm of 2 and cooled to 0°C to produce runoff (empty) ribosomes. Cells were harvested and frozen in liquid nitrogen. Frozen S. aureus cells were resuspended in TMKPL buffer [10 mM Tris, pH 8.2, with acetic acid, 14 mM Mg(OAc)₂, 60 mM KCl, 1 mM dithiothreitol, 1 mM phenylmethylsulfonyl fluoride, and 100 units/ml lysostaphin] at 4°C. The cell suspension was lysed by three consecutive passes through an EmulsiFlex-C5 microfluidizer (Avestin, Ottawa, Canada) cell, followed by the addition of DNase I to a final concentration of 1 unit/ml of the cell extract. The supernatant was centrifuged at 30,000 × g for 20 min at 4°C. The upper three-fourths of the resulting supernatant was mixed with 1.1 M sucrose (0.24 ml per ml of supernatant) and was centrifuged again at 30,000 × g for 2 h at 4°C. The cleared supernatant was loaded on a 1.1 M sucrose cushion and was centrifuged at 100,000 × g for 16 h. The ribosome pellet was suspended in TMK buffer and dialyzed for 2 h at 4°C against the same buffer, frozen in liquid nitrogen, and then stored at −80°C. Ribosomes (70S) were prepared from S. aureus (Lin^r Erm^r) in a fashion similar to S. aureus wild-type ribosomes except that the cells were grown in medium with erythromycin (50 μg/ml) to ensure a high level of A2058 methylation.

Translation inhibition assays.To test inhibition by Rχ-01 compounds of protein synthesis, we developed several prokaryotic in vitro translation-only assays. We developed these assays by modifying a previously described transcription/translation assay fueled with an E. coli S30 extract (Promega part number L1020) (35a). The translation-only assay uses purified S. aureus 70S ribosomes (20 nM final concentration) in TMK buffer (10 mM Tris-HCl, pH 7.4, 6 mM MgCl, 60 mM KCl, 1 mM dithiothreitol), various amounts of S100 extracts from different bacterial sources, Promega amino acid mix (0.1 mM final concentration), 3 μl of Promega S30 premix, and 200 to 800 nM (final concentration) of an in vitro-transcribed mRNA encoding firefly luciferase. The final volume of each translation reaction mixture was 10 μl. All compounds were tested in duplicate for translation inhibition, and all assays included both positive and negative controls to measure translation, either in the absence of a compound or in the presence of an antibiotic known to predictably and reproducibly inhibit translation. A Victor2V Multilabel Reader (Perkin Elmer) was used to read luminescence. To obtain compounds selective for prokaryotic ribosomes, we also tested the compounds to inhibit translation fueled by a rabbit reticulocyte lysate (nuclease-treated) system (Promega part number L4960), following the protocol in Promega's technical manual 232 (35b) and using mRNA (200 to 800 nM final concentration) encoding firefly luciferase. Fifty percent inhibitory concentrations (IC₅₀s) were calculated using MDL Assay Explorer with a one-site competition model of binding.

Competitive binding studies.70S ribosomes from S. aureus ATCC 29213 were incubated in ribosomal buffer (10 mM HEPES-KOH, pH 7.8, 10 mM MgOAc, 60 mM NH₄Cl, 6 mM mercaptoethanol) in the presence of either [³H]chloramphenicol or [³H]puromycin and increasing concentrations of unlabeled control antibiotics or Rχ-01 compounds to assess the binding competition. Ribosomes were separated from unbound compounds by spin column chromatography using Bio-Gel P-30 from Bio-Rad equilibrated with TMK buffer. The degree of radioactive chloramphenicol or puromycin displacement was quantified via scintillation counting (53). The apparent IC₅₀ was defined as the concentration of compound that displaced 50% of the bound chloramphenicol under fixed nonequilibrium conditions and was calculated using Prism V4 (GraphPad Software Inc.). Etamycin, griseoviridin, and tylosin were used to validate the method (data not shown).

RNA footprinting.To determine whether Rχ-01 compounds protect 23S rRNA bases from chemical modification, 70S ribosomes (50 to 200 nM) were incubated for 10 min at 37°C, followed by 10 min at 20°C in 50 μl of the corresponding modification buffers (46) containing 0.1 to 1 mM antibiotics. 1-Cyclohexyl-3-(2-morpholinoethyl) carbodiimide metho-p-toluene sulfonate (CMCT) and dimethyl sulfate modifications were carried out at 37°C or at room temperature, as described previously (31). Primer extension was performed in the presence of [α-³³P]TTP according to the method of Stern et al. (46) with a set of primers specific for either E. coli or S. aureus 23S rRNA, depending on the source of ribosomes used in the experiment. Dried gels were exposed and bands were quantified with a “Storm” scanner (Molecular Dynamics).

Translational-accuracy assay.To perform translational-accuracy assays, E. coli strains containing plasmids to test translational accuracy (48) were incubated with sub-MIC concentrations of test compounds. In these strains, the level of β-galactosidase activity is dependent on the abilities of the test compounds to promote either a +1/−1 frameshift event or a stop codon readthrough. Each experimental point was generated by averaging the results of three separate assays measured in duplicate (33). E. coli MC245 containing one of four different lacZ plasmids was used to determine the level of translational inaccuracy caused by Rχ-01 compounds. Two of the constructs allowed us to test for either +1 frameshifting or −1 frameshifting, while the other two constructs allowed us to test for stop codon readthrough of either UAG or UGA stop codons. Because the levels of β-galactosidase produced from each of the reporter plasmid constructs varied in the absence of antibiotic, the values for the zero drug controls were set at one to allow direct comparison between constructs. Reporter cells were grown to an optical density at 600 nm of 0.3 in the presence (12 to 14 h) or in the absence (4 h) of subinhibitory drug concentrations (48).

Rχ-01 compounds overcome ribosome-based linezolid resistance.We compared the ability of the Rχ-01 family of oxazolidinones to inhibit bacterial protein synthesis to that of linezolid in a translation-only assay using ribosomes isolated from two different sources: S. aureus ATCC 29213 (wild type) or S. aureus A7820 (Lin^r Erm^r). S. aureus A7820 has a plasmid carrying ermC methylase, as well as a G2576U mutation in each allele of 23S rRNA. Most linezolid resistance in staphylococci and enterococci isolated in clinical settings is due to this G2576U mutation (22, 36, 38, 39, 49, 51). Our results indicated that Rχ-01 compounds inhibit translation fueled by S. aureus wild-type ribosomes up to ≥45-fold better than linezolid (Table 1). When translation was mediated by Lin^r Erm^rS. aureus ribosomes, there was a 10-fold increase in the IC₅₀ for linezolid and only at most a ≤4-fold increase for Rχ-01 compounds compared to wild-type values. Furthermore, the abilities of the Rχ-01 compounds to bind tightly enough to ribosomes to overcome ribosome-based linezolid resistance do not come at the expense of selectivity. Translation assays showed that most Rχ-01 compounds are at least 100-fold less active in inhibiting translation in rabbit reticulocytes than in S. aureus ribosomes. Thus, Rχ-01 compounds not only bind more tightly to ribosomes than linezolid, they also display a selectivity ratio comparable to that of linezolid (Table 1).

Antimicrobial activities of Rχ-01 compounds.We determined MICs on linezolid-susceptible and -resistant S. aureus and enterococcal clinical isolates bearing the G2576U mutation. Our study showed that all members of the Rχ-01 family were equally or more effective than linezolid in inhibiting the growth of S. aureus and E. faecalis strains susceptible to linezolid (S. aureus QC [Table 2]). When potency at the ribosomal target was assessed by in vitro translation assays, most members of the Rχ-01 family of novel oxazolidinones, in contrast to linezolid, chloramphenicol, and florfenicol, overcame resistance linked to the G2576U mutation. Although the IC₅₀s in translation for S. aureus Lin^r Erm^r ribosomes (Table 1) did not always mirror the ranking of Rχ-01 compounds according to their antibacterial activities compared to linezolid-resistant strains (Table 2), in every case the Rχ-01 compounds were more potent than linezolid against linezolid-resistant isolates, reflecting their tighter binding to the ribosome. Studies detailing the in vitro potencies versus different community- and hospital-acquired pathogens have already been published (26).

Rχ-01 compounds protect nucleotide U2585 of 23S rRNA.To determine the binding target of Rχ-01 compounds, we also made use of chemical modifications of 23S rRNA. Our study showed that nucleotide U2585 is protected from CMCT modification in the presence of chloramphenicol but not in the presence of linezolid, despite overlapping binding sites (37) (Fig. 2B). Interestingly, nucleotide U2585 is protected from CMCT modification by the binding of Rχ-01 compounds (Fig. 2A and B). The level of U2585 protection attained in the presence of Rχ-01 compounds (40 to 60%) is comparable to that of chloramphenicol (50%) and lower than that of florfenicol (80%). A similar pattern of protection was observed for other compounds that bind to the 50S ribosomal A site (e.g., tiamulin and streptogramin A) (data not shown). These results are in good agreement with the crystallographic structures obtained for various Rχ-01 compounds bound to the 50S ribosomal subunits of H. marismortui (21) and confirm that Rχ-01 compounds interact with the ribosome in a manner notably different from that of conventional oxazolidinones.

• Open in new tab
• Download powerpoint

FIG. 2.

(A) CMCT modification of 23S rRNA in 70S S. aureus ribosomes in the presence of several Rχ-01 compounds at two different concentrations (conc.) (0.1 mM and 1 mM). The results from primer extension shown in the gel correspond to the nucleotide region from 2550 to 2610 in 23S rRNA. (B) Protection of U2585 by Rχ-01 and control compounds. The bar graph was produced by quantifying the relative band intensities of U2585 footprinting in the presence of CMCT from a gel similar to the one shown in panel A.

Determining the structure of linezolid bound to H. marismortui 50S required crystals to be soaked in >4 mM linezolid, a concentration that is at least 40 times higher than the concentration used in the chemical protection experiments. The high concentration needed to obtain the structure of linezolid gives additional support to the inability of linezolid, in contrast to Rχ-01 compounds, to protect U2585 at submillimolar concentrations. Figure 3A shows a rendering of the X-ray structure of linezolid bound to the 50S subunit of H. marismortui (20) and highlights the position of U2585 (E. coli numbering) with respect to the oxazolidinone binding site. Figure 3B shows the positions of linezolid and U2585 with respect to the CCA ends of A- and P-site tRNAs, showing that oxazolidinones, when bound to this pocket, are able to interfere with A-site tRNA positioning.

• Open in new tab
• Download powerpoint

FIG. 3.

Longitudinal cut of the 50S subunit of H. marismortui showing the structure of linezolid bound to the vicinity of the A site (20). (A) Orientation of U2585 (blue) with respect to linezolid (shown in dark yellow). (B) Position of linezolid with respect to the CCA ends of the A site (surface) and P site (sticks) of tRNA (yellow, C; green, A) (42). The figures were made in VMD (19) and rendered with Raster3D (30).

Rχ-01 compounds have high binding affinity for the ribosome.Structural studies of Deinococcus radiodurans 50S have shown chloramphenicol bound at the peptidyl transferase center (40). Nevertheless, subsequent X-ray studies using 50S from H. marismortui found a second chloramphenicol binding site at the entrance of the peptide exit tunnel (18). Therefore, we investigated the abilities of Rχ-01 compounds to displace puromycin, in addition to chloramphenicol. Puromycin is the prototypical A-site antibiotic and has been shown by X-ray crystallography to bind in a fashion similar to those of the A site of H. marismortui (18, 32) and D. radiodurans ribosomes (3).

Because Rχ-01 compounds and chloramphenicol protect U2585 from chemical modification by carbodiimide at about the same level, we assessed if Rχ-01 compounds could displace the A-site inhibitor chloramphenicol or puromycin from its ribosome binding site. The chloramphenicol displacement assay showed that members of the Rχ-01 family of compounds displace chloramphenicol with apparent IC₅₀s 20-fold lower than that for linezolid (Fig. 4A shows Rx-01_007 and Rx-01_423). The abilities of Rχ-01 compounds to displace chloramphenicol and to protect nucleotide U2585 are well correlated, as shown in Fig. 2. Equally, the inability of linezolid to protect U2585 is reflected in its poor ability to displace chloramphenicol from the ribosome. Similar experiments using radiolabeled puromycin instead of chloramphenicol showed that Rx-01_149 was able to displace puromycin with an IC₅₀ 100-fold lower than that for linezolid (Fig. 4B). These results suggest that Rχ-01 compounds bind to the ribosomal A site and confirm that they do so with higher affinity than linezolid.

• Open in new tab
• Download powerpoint

FIG. 4.

Displacement of [³H]chloramphenicol (A) or [³H]puromycin (B) from S. aureus (wild-type) 70S ribosomal complexes by Rχ-01 compounds and controls. The apparent IC₅₀ is defined as the concentration of the compound that displaces 50% of the bound [³H]chloramphenicol or [³H]puromycin under fixed nonequilibrium conditions. The error bars indicate standard deviations.

Rχ-01 oxazolidinones cause translational inaccuracy.Linezolid, chloramphenicol, and other inhibitors that bind to the 50S ribosomal subunit promote frameshifting and readthrough of stop codons (28, 48). These effects on translational fidelity may play a significant role in the mechanisms of action of these compounds. Therefore, we investigated whether the Rχ-01 compounds were able to promote translational inaccuracy in a fashion similar to that of linezolid. We found that Rχ-01 compounds were able to promote stop codon readthrough two- to fivefold more efficiently than linezolid (Fig. 5A). Specifically, Rx-01_423-induced stop codon readthrough is five-fold more efficient than that of linezolid, while Rx-01_667 and Rx-01_413 were three- and twofold more efficient than linezolid. The Rχ-01 compounds Rx-01_413 and Rx-01_423 promoted translational inaccuracy as measured by the promotion of a −1 frameshifting event (Fig. 5B) at extents that were approximately equal to that of linezolid. Rx-01_413 and Rx-01_423 were able to stimulate +1 frameshifting about twofold more efficiently than linezolid, while Rx-01_667 was able to promote both −1 and +1 frameshifting about threefold more than linezolid (Fig. 5B), suggesting that this effect could contribute to their antibacterial activities.

• Open in new tab
• Download powerpoint

FIG. 5.

(A) Production of β-galactosidase by readthrough of stop codons. E. coli strains (48) carrying reporter plasmids with either stop codon at the N terminus of the lacZ gene were grown in the presence of subinhibitory concentrations of gentamicin (1.5 μg/ml), linezolid (8 μg/ml), Rx-01_413 (0.5 μg/ml), Rx-01_423 (2 μg/ml), and Rx-01_667 (1 μg/ml). (B) Production of β-galactosidase is dependent on frameshifting events. E. coli strains carrying reporter plasmids in which the reading frame of the lacZ gene contained either a +1 or a −1 frameshift close to the N terminus were grown as described for panel A. Frameshifting and stop codon readthrough values in the absence of antibiotics were set at 1 to allow the relative effects of all antibiotics tested to be compared. The stop codon readthrough and frameshift values obtained for linezolid and for the gentamicin control in this assay were comparable to those obtained by Thompson et al. (48). The error bars indicate standard deviations.