ABSTRACT
Antibiotic-resistant bacteria are widespread and pose a growing threat to human health. New antibiotics acting by novel mechanisms of action are needed to address this challenge. The bacterial methionyl-tRNA synthetase (MetRS) enzyme is essential for protein synthesis, and the type found in Gram-positive bacteria is substantially different from its counterpart found in the mammalian cytoplasm. Both previously published and new selective inhibitors were shown to be highly active against Gram-positive bacteria with MICs of ≤1.3 μg/ml against Staphylococcus, Enterococcus, and Streptococcus strains. Incorporation of radioactive precursors demonstrated that the mechanism of activity was due to the inhibition of protein synthesis. Little activity against Gram-negative bacteria was observed, consistent with the fact that Gram-negative bacterial species contain a different type of MetRS enzyme. The ratio of the MIC to the minimum bactericidal concentration (MBC) was consistent with a bacteriostatic mechanism. The level of protein binding of the compounds was high (>95%), and this translated to a substantial increase in MICs when the compounds were tested in the presence of serum. Despite this, the compounds were very active when they were tested in a Staphylococcus aureus murine thigh infection model. Compounds 1717 and 2144, given by oral gavage, resulted in 3- to 4-log decreases in the bacterial load compared to that in vehicle-treated mice, which was comparable to the results observed with the comparator drugs, vancomycin and linezolid. In summary, the research describes MetRS inhibitors with oral bioavailability that represent a class of compounds acting by a novel mechanism with excellent potential for clinical development.
INTRODUCTION
Gram-positive bacteria, such as Staphylococcus, Streptococcus, and Enterococcus, are major human pathogens responsible for a myriad of clinical syndromes. Antibiotic-resistant strains, such as methicillin-resistant Staphylococcus aureus (MRSA) and vancomycin-resistant Enterococcus (VRE) strains, are widespread, and the effectiveness of the available antibiotics against such strains is limited. Concern about the diminishing availability of effective antibiotics has led to urgent calls for the development of new antibiotics (1). Targeting the prokaryotic protein synthesis machinery has been a highly successful strategy for the development of antibiotics. Aminoglycosides, tetracyclines, macrolides, ketolides, and oxazolidinones are major classes of antibiotics that all interfere with bacterial protein translation. Inhibition of tRNA synthetases represents another possible approach to target prokaryotic protein translation. The widely used antibiotic mupirocin works by inhibiting the bacterial isoleucyl-tRNA synthetase (2). Mupirocin is used as an ointment to treat or decolonize patients with cutaneous infections due to Staphylococcus or Streptococcus; however, its use is limited to the topical route of administration. Another bacterial tRNA synthetase inhibitor, a boron-containing compound targeting the bacterial leucyl-tRNA synthetase (GSK2251052), made it to phase 2 trials for the treatment of Gram-negative bacterial infections (3). Its development was discontinued, unfortunately, due to the high rates of resistance that occurred during treatment, which may have been related to the targeting of the editing domain of the enzyme rather than the catalytic domain. Investigators at GlaxoSmithKline reported on inhibitors of the S. aureus methionyl-tRNA synthetase (MetRS) over a decade ago (4–7). These inhibitors had excellent antibiotic potency but poor oral (p.o.) bioavailability that restricted their development (pre-new drug application) to topical use for skin infections and to oral use for Clostridium difficile infections, where oral absorption is not needed (8, 9). The research presented in this report also focuses on MetRS inhibitors, building on compounds that are being developed as antiprotozoan chemotherapies (10–13). The compounds have high selectivity (>1,000-fold) for Trypanosoma brucei cells over mammalian cell lines (14). Changes to the molecules have led to improved oral bioavailability and pharmacokinetic properties (14), thus making them better candidates for antibiotic development, as will be described.
With respect to the target, bacteria and all living organisms contain a complement of tRNA synthetases that are responsible for charging tRNAs with their corresponding amino acids for subsequent delivery to the ribosome. tRNA synthetases, including MetRS, catalyze a two-step reaction, as follows:
RESULTS
MetRS inhibitors and lead optimization.The structures and properties of the compounds under study in the experiments described here are shown in Fig. 1. The procedures for the synthesis of compounds 1312, 1575, 1614, and 1717 were published previously (10, 14). The procedures for the synthesis of new compounds 1962, 2062, 2093, 2114, and 2144 are described in the supplemental material. The starting point for these investigations was the aminoquinolone scaffold, exemplified by compound 1312 (Fig. 1). In separate research to develop MetRS inhibitors as antiprotozoan drugs, our group introduced changes to the molecules with the goals of improving oral bioavailability while retaining potent activity against the MetRS target. The evolution of these compounds included changing of the aminoquinolone group to a fluorinated imidazopyridine (e.g., compound 1614), which improved the oral bioavailability in mice from <10% for compound 1312 to ∼40% for compound 1614 (14). Subsequent changes to the linker region reported previously (13) and in this paper have further improved the potency and pharmacological properties of the series. The results of testing of the activities of the MetRS inhibitors against recombinant S. aureus MetRS, bacterial cultures, and mammalian cells are shown in Table 1. All compounds tested had 50% inhibitory concentrations (IC50s) for the S. aureus MetRS below the level of sensitivity of the assay (25 nM). The MetRS inhibitors had potent activity against a variety of Gram-positive bacterial strains but essentially no activity against Gram-negative bacteria (Escherichia coli and Pseudomonas aeruginosa). Specifically, MIC values below 0.3 μg/ml were measured against strains of S. aureus (including methicillin-sensitive S. aureus [MSSA], MRSA, and vancomycin-intermediate S. aureus [VISA] strains), Staphylococcus epidermidis, Enterococcus faecalis, and Enterococcus faecium (including vancomycin-susceptible Enterococcus [VSE] and VRE strains). The compounds with the lowest MICs were 1717, 2093, and 2144, which were >10 times more potent than the control drugs vancomycin and linezolid against many strains. These compounds were the subject of further investigations, discussed below. Higher MICs against Streptococcus pyogenes were seen, and no activity against Streptococcus pneumoniae was observed. The selectivity for staphylococci versus mammalian cells (determined by comparison of the MIC to the 50% cytotoxic concentration [CC50]) was at least 35-fold for the three most potent compounds, compounds 1717, 2093, and 2144.
Chemical structures of MetRS inhibitors.
Assay results for representative MetRS inhibitorsa
Microbiological characterization of selected compounds. (i) Macromolecular synthesis assays.In order to verify that the compounds were acting by the expected mechanism of action, radioisotope incorporation assays were performed (Fig. 2). Incorporation of the amino acid [3H]lysine was inhibited by the MetRS inhibitors (compounds 1717, 2093, and 2144), consistent with inhibition of protein synthesis. The findings were similar to those seen with linezolid (Fig. 2A), which is known to inhibit protein synthesis by interfering with the bacterial ribosome (20). In contrast, the MetRS inhibitors had less of an effect on both the incorporation of [3H]uridine (a measure of RNA synthesis) and the incorporation of [3H]thymidine (a measure of DNA synthesis) (21). Ciprofloxacin showed selective inhibition of DNA synthesis (Fig. 2B), consistent with its mechanism as an inhibitor of DNA topoisomerases. Finally, rifampin showed selective inhibition of RNA synthesis (slightly greater than its inhibition of protein synthesis) (Fig. 2C), consistent with its mechanism as an inhibitor of bacterial RNA polymerase.
Macromolecular synthesis experiments. The level of incorporation of radiolabeled precursors into S. aureus (ATCC 29213) was measured in 30-min incubations in the presence of established antibiotics or MetRS inhibitors. Dashed vertical lines, MICs.
(ii) Activity against permeable E. coli strains.The purpose of these experiments was to determine if the nonsusceptibility of Gram-negative bacterial strains (e.g., E. coli ATCC 25922, shown in Table 2) was due to the inability of the MetRS inhibitors to penetrate the Gram-negative bacterial cell wall. Mutant MB4902 is an outer membrane-permeable E. coli strain and showed no greater susceptibility to three MetRS inhibitors (compounds 1717, 2093, and 2144) than the wild-type E. coli strain (MB4827). Similarly, efflux-negative strain MB5747 showed no increased susceptibility to the MetRS inhibitors, nor did the mutant containing both mutations (MB5746). The control drug, novobiocin, had increased activity against the hyperpermeable E. coli strains, as has been previously reported (22, 23).
MICs of selected MetRS inhibitors against E. coli strains with increased permeability
(iii) MIC/MBC.Measurements of minimum bactericidal concentrations (MBCs) were done with S. aureus strain ATCC 29213 (Table 3). The MBC is defined as the drug concentration that reduces bacterial growth by ≥99.9%. Compounds exhibiting an MBC/MIC ratio of ≤4 are generally considered bactericidal, while an MBC/MIC ratio of >4 is considered bacteriostatic (24). The data indicate that compounds 1717, 2093, and 2144 have bacteriostatic activity similar to that of linezolid.
MICs and MBCs against S. aureus (ATCC 29213)
(iv) Resistance frequency rates.The propensity for S. aureus (ATCC 29213) to develop resistance to MetRS inhibitors was also studied (Table 4). This was done by plating high numbers (3.8 × 109 CFUs in experiment 1 and 5.5 × 109 CFUs in experiment 2) of S. aureus on tryptic soy agar (TSA) plates impregnated with one of the compounds at a concentration of 4× or 8× the MIC and incubating for 72 h. The resistance frequency rates for compounds 1717, 2093, and 2144 at 8× the MIC were in the average range of 2 × 10−8 to 4 × 10−9. These rates are comparable to those for the test drug novobiocin but higher than the rates found for ciprofloxacin or linezolid.
Frequency of spontaneous resistance for S. aureus (ATCC 29213)a
(v) Serum shift and protein binding assays.Serum shift assays were done to analyze the impact of protein binding on the MICs (Table 5). The MIC shifts in the presence of 50% human serum ranged from 16-fold to 128-fold for the MetRS inhibitors, which is consistent with a high level of protein binding (e.g., 95.4% for compound 1717). Although the shifts were much higher than the shift for vancomycin (2-fold), the absolute MICs of some compounds in serum (e.g., compounds 1717 and 2144) were still comparable to the MIC of vancomycin (range, 1 to 4 μg/ml).
MIC values for S. aureus (ATCC 29213) grown in MHB or MHB with 50% human seruma
Metabolic stability studies.The compounds were incubated with murine or human liver microsomes to evaluate their stability to hepatic metabolic enzymes (Table 6). Metabolism rates were similar between mouse and human microsomes for individual MetRS inhibitors. Compound 2093 demonstrated the most rapid metabolism (3.2 min in mouse microsomes), whereas compound 1962 demonstrated the highest metabolic stability (27.0 min in mouse microsomes). The drug linezolid was more metabolically stable, with half-lives of >145 min in both mouse and human liver microsomes.
Liver microsome stability half-lives
Pharmacokinetics studies.Selected MetRS inhibitors (compounds 1614, 1717, 2093, and 2144) and linezolid were administered to mice in single oral or intravenous (i.v.) doses, and tail blood was sampled at time intervals out to 24 h to assess blood exposure (see Table S2 in the supplemental material). The terminal half-lives of the MetRS inhibitors in blood ranged from 17 to 58 min (whereas it was 35 min for linezolid). Clearance ranged from 22 to 63 ml/min/kg (whereas it was 20 ml/min/kg for linezolid). The maximum blood concentration (Cmax) following oral dosing ranged from 0.39 to 9.3 μM (whereas it was 13.7 μM for linezolid). The area under the concentration-time curve (AUC) in blood following oral dosing ranged from 117 to 615 min · μM (whereas it was 1,707 min · μM for linezolid). Finally, the apparent oral bioavailability ranged from 24 to 46% (whereas it was 94% for linezolid).
Efficacy studies in mice.Selected MetRS inhibitors were tested for in vivo efficacy in the model of S. aureus thigh infection in the neutropenic mouse. Compounds 1717 and 2144 resulted in an ∼3- to 4-log decrease in the number of CFU compared to that for the vehicle group, similar to the results for vancomycin and linezolid (Fig. 3). Note that the drop was below the stasis level, which was determined by harvesting of a group of mice at 1 h postinfection.
Efficacy of MetRS inhibitors in a neutropenic mouse model of S. aureus thigh infection. Error bars are SEMs. The stasis level was determined from untreated mice sacrificed at 1 h postinfection (p.i.). Vanco, vancomycin.
Protein sequence analysis.Using coordinates from the T. brucei MetRS complex with inhibitor 1312 (PDB accession number 4EG5 ), the residues in the binding site of the inhibitor were aligned for various species (Table 7). S. aureus MetRS (strain USA300; UniProtKB accession number A0A0H2XID2 ) has an extremely high degree of sequence conservation with the Trypanosoma brucei MetRS, with 22 of 25 amino acids being identical (and with 23 amino acids potentially being identical, since position 456 could be either an Leu or an His, but the sequence in the model is ambiguous due to the loop length). This confirms that many inhibitors of the T. brucei MetRS will likely inhibit the S. aureus MetRS. We also compared the sequence of the the human mitochondrial MetRS (UniProtKB accession number Q96GW9 ) with the S. aureus MetRS sequence and identified four different amino acid residues (at positions 249, 291, 470, and 471). Three of these changes occur in pocket q, which binds the quinolone moiety of compound 1312.
Inhibitor binding site amino acid residues for MetRS enzymes
DISCUSSION
The essential enzyme methionyl-tRNA synthetase was targeted for the discovery of antibiotics with activity against Gram-positive bacteria. The research capitalizes on progress to develop antimicrobial agents with activity against pathogenic protozoa, including Trypanosoma brucei and Giardia intestinalis (10–14, 25). In particular, the challenges associated with the poor oral bioavailability observed with early aminoquinolone compounds, such as compound 1312 (10, 22, 26), were significantly improved with the fluoroimidazopyridine derivatives (compounds 1614 to 2144; Fig. 1) (14). Compounds with the aminoquinolone scaffold were optimized for activity against the T. brucei MetRS (a type 1 enzyme). The comparison of the protein sequences of the MetRS enzymes of T. brucei and S. aureus showed identity of 22 of 25 amino acid residues in the inhibitor binding site (Table 7), suggesting that cross activity from T. brucei to S. aureus was likely. In fact, all the compounds tested for inhibitory activity against the recombinant S. aureus MetRS enzyme (Table 1) demonstrated IC50s below 25 nM, the sensitivity limit of the assay. Further titration below this concentration was not possible with the applied methods due to the need for 25 nM enzyme to give a suitable signal for measurement (see Materials and Methods). The assays against live bacterial cultures demonstrated the potency of the MetRS inhibitors in the sub-microgram-per-milliliter range against Staphylococcus and Enterococcus species and lower levels of potency against streptococci (Table 1). Of the compounds with the linear linker structure (e.g., compounds 1312, 1575, 1614, and 1717), compound 1717 was the most potent, with a MIC of 0.16 μg/ml. This was the compound with the most potent activity against T. brucei cultures (14). Compounds with the ring system in the linker region of the scaffold (e.g., compounds 1962 to 2144), including compounds 2093 and 2144, with MICs of 0.04 and 0.02 μg/ml for MSSA ATCC 29213, respectively, also had potent antistaphylococcal activity (Table 1). Compound 2093 contains a single ring in the linker (an imidazole-2-one), whereas compound 2144 contains a fused imidazo[1,2-a]pyridine ring system in the linker region.
The spectrum of activity of the MetRS inhibitors against ATCC strains of pathogenic Gram-positive and Gram-negative bacteria was explored. As was predicted, the antibiotic activity was restricted to bacteria dependent upon the type 1 MetRS enzyme, i.e., Gram-positive bacteria. Gram-negative bacteria (i.e., Escherichia coli and Pseudomonas aeruginosa), which are known to contain the type 2 MetRS enzyme (17), were insensitive to all the tested compounds at the highest concentration tested of 10 μg/ml. The selectivity for Gram-positive organisms is potentially advantageous, in that the MetRS inhibitor developed as a drug will not add to the resistance of nontargeted Gram-negative bacteria. Sensitive Gram-positive strains were S. aureus, Enterococcus faecium, Enterococcus faecalis, and Staphylococcus epidermidis. Furthermore, these included drug-resistant strains, such as MRSA, VISA, and VRE strains, whose mechanisms of resistance to semisynthetic penicillins and glycopeptide antibiotics are unrelated to the cellular processes inhibited by the MetRS inhibitors. The MICs of the MetRS inhibitors for S. pyogenes (ATCC 19615) were higher than those for S. aureus and Enterococcus strains, which we ascribe to the need to grow S. pyogenes in medium containing lysed blood. We have already shown the effects of plasma protein binding on the MICs of the compounds (Table 5), and we suspect the effect of the addition of laked horse blood to be similar. With the shift observed with blood, the MIC for compound 2144 (1.3 μg/ml) was about the same as the MICs observed for vancomycin and linezolid (0.63 and 1.3 μg/ml, respectively). The Gram-positive coccus Streptococcus pneumoniae (ATCC 49619) was resistant to the MetRS inhibitors (MICs > 10 μg/ml). This finding is consistent with previous reports that ∼45% of S. pneumoniae strains are resistant to type 1 MetRS inhibitors due to the presence of a second (type 2) MetRS inhibitor in the genome (16). It is likely that MetRS inhibitors would need to be used with caution for the treatment of pneumonia or other clinical syndromes in which S. pneumoniae is commonly found, at least until cultures rule out the possibility that S. pneumoniae is the cause of the infection. Future studies will investigate a broader collection of S. pneumoniae isolates to assess the MIC range against this pathogen. The issue of a secondary MetRS gene has not been described in other Gram-positive bacteria, so this is unlikely to be a broader concern. We expect that MetRS inhibitors will be active against many other bacteria containing the type 1 MetRS enzyme, including species of Clostridium, Corynebacterium, Bacillus, Propionibacterium, Actinomyces, and others (16). Various species of these are, of course, pathogenic in humans, and their susceptibility will be tested in future studies. An exception to the rule for Gram-positive bacteria mentioned above is Brucella (a Gram-negative rod), which is known to contain a type 1 MetRS and is susceptible to MetRS inhibitors (27).
In order to address the question about the target of action in living bacteria, macromolecular synthesis assays were run with MetRS inhibitors and various control drugs (Fig. 2). As was expected, MetRS inhibitors resulted in rapid dose-dependent decreases in the uptake of a radiolabeled amino acid (Lys), consistent with the disruption of protein synthesis. The changes were similar to those seen with the protein synthesis inhibitor linezolid. At the same time, RNA and DNA synthesis was unaffected by MetRS inhibitors (compounds 2093 and 2144) until concentrations above the MIC were used, while the control drugs rifampin and ciprofloxacin caused selective inhibition of these pathways, respectively, in the anticipated manner. Compound 1717 showed mild suppression (20 to 40%) of RNA and DNA at the MIC, which could suggest an undefined secondary target or perhaps kinetics of binding to MetRS different from those for compounds 2093 and 2144. These studies provide assurance that the compounds are likely to mediate their antibiotic effects through inhibition of the MetRS target in vivo.
In a similar vein, the activities of selected MetRS inhibitors against strains of E. coli with defects in cell wall permeability and/or efflux were tested (Table 2). The purpose of these experiments was to show that the resistance of E. coli was not due to the exclusion of the MetRS inhibitors by the Gram-negative bacterial cell wall or efflux but, rather, was due to inherent resistance. The findings that the cell-permeable strains were resistant to the three most potent MetRS inhibitors (compounds 1717, 2093, and 2144) are consistent with the understanding that E. coli contains a type 2 MetRS enzyme (17) which is not inhibited by the compounds under development. Furthermore, it indicates that off-target mechanisms of action are not at play, at least with this species of bacteria.
The minimum bactericidal concentrations of compounds 1717, 2093, and 2144 against the S. aureus ATCC 29213 strain were determined. The MBC/MIC ratio for these three compounds was between 16 and 32. A ratio of 4 or less is considered bactericidal (28); thus, the MetRS inhibitors would be considered bacteriostatic against this strain of S. aureus. An MBC/MIC ratio of 64 was observed with the clinical drug linezolid (which is known to be bacteriostatic), whereas the ratio for vancomycin was 4, consistent with its bactericidal mechanism.
The rates of S. aureus (ATCC 29213) resistance to MetRS inhibitors were determined on agar plates containing MetRS inhibitors at concentrations 4 or 8 times their MICs. The resistance frequency rates for MetRS inhibitors (at 8× MIC) were between 2 × 10−8 and 4 × 10−9, which are higher than those observed for ciprofloxacin and linezolid (Table 4). Resistance frequency rates in the range of 10−6 to 10−9 are indicative of a single drug target within the cell (28), which is consistent with the understanding of the mechanisms of action of these compounds. Resistance frequency rates for drugs such as rifampin are even higher (2 × 10−7) (28), but drugs with such resistance frequency rates are generally used in combination with other drugs to avoid generating resistance. Vancomycin and linezolid are known to have low resistance frequency rates (<10−11), and along with this characteristic, relatively little resistance (at least by staphylococci) to these drugs has developed in the clinic. Further research will be necessary to find out if the rates of resistance to MetRS inhibitors are problematic for their clinical development as monotherapy agents. If the risk for resistance development appears to be high, then the development of the compounds for administration with a partner antibiotic may be an attractive option to mitigate the problem.
The MetRS inhibitors characterized in this report exhibited properties of high levels of protein binding (95 to 99.9%). The low unbound concentration of compounds translates to substantial effects when MICs are measured in the presence of serum (Table 5). Serum shifts ranging from 16-fold to 128-fold were observed with the series of compounds tested. For perspective, vancomycin demonstrates only about a 2-fold serum shift (Table 5), whereas fusidic acid is reported to have 97% protein binding and a 130-fold increase in the MIC for S. aureus in the presence of 50% serum (28). Due to the high potency of the MetRS inhibitors, the MICs of compounds 1717 and 2144 in the presence of 50% serum (1 and 4 μg/ml, respectively) were comparable to the MIC of vancomycin (2 μg/ml).
Incubation of the MetRS inhibitors with mouse or human liver microsomes showed variable rates of metabolism, although the half-lives were generally relatively short (<20 min for human microsomes and <10 min for mouse microsomes) (Table 6). The pharmacokinetic studies in mice showed clearance values for the MetRS inhibitors ranging from 18 to 63 ml/min/kg, whereas the clearance value for linezolid was 20 ml/min/kg. The fact that the clearance of MetRS inhibitors is similar to that of linezolid, despite the more rapid microsome metabolism, is likely attributable to the high level of plasma protein binding, which can protect compounds from liver cytochrome P450 metabolism (29). The oral bioavailability of the MetRS inhibitors ranged from 24 to 46%, which is substantially higher than that of the original aminoquinolone compounds, such as compound 1312 (oral bioavailability, <10%) (14), but lower than that of linezolid (94%). This difference in oral bioavailability is probably responsible for the AUC of linezolid (1707 min · μM) following oral dosing higher than that observed with the MetRS inhibitors (117 to 615 min · μM). As will be discussed below with the efficacy results, the combined properties of the compounds (particularly compounds 1717 and 2144) appear to be sufficient to clear bacteria from infected mice with an efficiency similar to that of vancomycin or linezolid.
The results of the efficacy experiments are very encouraging and indicate the prospects of developing MetRS inhibitors as antibiotics. The results of a pilot experiment (not shown) and two independent experiments showed the reproducibility of the S. aureus thigh infection model in mice made neutropenic by cyclophosphamide pretreatment. The model represents a soft tissue infection which resembles the disease process (skin and skin structure infection) to which the clinical development of the compounds would initially be targeted. Both compound 1717 and compound 2144 demonstrated a significant reduction of the bacterial load below the stasis level at least as effectively as the comparator drugs vancomycin and linezolid. The approximately 2-log reduction below stasis levels in a neutropenic mouse is noteworthy, in light of the bacteriostatic activity observed in vitro (Table 3). This shows that tissue levels at the site of infection were sufficient to substantially reduce the levels of bacteria even in the absence of neutrophils. Many bacteriostatic antibiotics (including linezolid) are widely and successfully used in the clinic, so the bacteriostatic characteristic of the MetRS inhibitors may not be a significant liability. Interestingly, compound 2093 was found to have weaker activity than the other MetRS inhibitors evaluated. The explanation probably relates to the particularly high level of protein binding of this compound, which presumably reduces the levels of unbound compound below the threshold needed to exceed the MBC at the site of infection. Future dose-response experiments will help us determine the relative potency of compounds 1717 and 2144 compared to each other and to those of additional MetRS inhibitors under development.
To date, we have observed no apparent side effects of the MetRS inhibitors given to mice. In this study, uninfected mice receiving a single dose (50 mg/kg of body weight) of compounds for pharmacokinetic analysis had no acute reactions during the 24-h observation period. In a previous study, compounds 1614 and 1717 were administered to mice that had been infected with T. brucei for 10 days at 50 mg/kg p.o. twice per day with no deleterious effects on weight, grooming, or body condition (14). The cytotoxicity of the MetRS inhibitors against mammalian cells was low (Table 1). For example, the ratios of the CC50 to the MIC for compounds 2093 and 2144 were >500, demonstrating a wide therapeutic window. A potential toxicity concern for the MetRS inhibitors is inhibition of the mammalian mitochondrial MetRS enzyme, which bears close homology to the S. aureus MetRS (Table 7). Manifestations of this potential toxicity have not been evident with in vitro cytotoxicity testing (by the 48-h assay against lymphocyte and hepatocyte cell lines) or in mice, as described above. Many antibiotics acting as protein synthesis inhibitors are known to inhibit mitochondrial protein synthesis as an off-target effect (30, 31). These include widely used drugs, such as tetracycline, erythromycin, aminoglycosides, and linezolid. Instead of directly affecting mitochondrial oxidative phosphorylation, these drugs interfere with mitochondrial biogenesis and are relatively slow to result in clinical problems, often with tissue-specific toxicity, depending on the particular drug. Linezolid, for example, is known to cause hematologic disturbances, peripheral neuropathy, and metabolic acidosis when it is administered for more than a 28-day period (32). The fact that these side effects are slow to manifest makes them more tolerable for antibiotics since treatment courses are typically relatively short (<10 days). Studies of the effects of MetRS inhibitors on mammalian mitochondrial function will be part of future investigations.
In summary, with the aid of structure-based drug design (14), new MetRS inhibitors with potent and broad-spectrum activity against Gram-positive bacteria have been developed. Macromolecule labeling studies demonstrated the inhibition of protein synthesis, consistent with the designed mechanism of action. As with other protein synthesis inhibitors, such as oxazolidinones, tetracyclines, and lincosamides, the MetRS inhibitors have bacteriostatic properties against S. aureusin vitro. The compounds are highly protein bound, which may help sustain plasma levels in vivo by limiting availability to cytochrome P450 metabolism. At least two MetRS inhibitors displayed activity comparable to that of linezolid in the neutropenic mouse thigh infection model, which indicates that the free fraction of compound is sufficient to inhibit bacterial growth. In fact, the bacterial load decreased by 1 to 2 logs below the stasis level, indicating that the in vivo activity was not just bacteriostatic but also bactericidal. These studies and previous reports (10, 14) have shown that the MetRS inhibitors have little in vitro toxicity and appear to be well tolerated when dosed in mice for up to 10 days. Additional preclinical toxicology studies are planned to further investigate the potential for adverse effects from inhibition of the mitochondrial MetRS or other off-target activities. In total, these MetRS inhibitors with oral bioavailability represent a class of compounds acting by a novel mechanism with excellent potential for clinical development.
MATERIALS AND METHODS
Media and culture conditions.Mueller-Hinton broth (MHB), cation-adjusted Mueller-Hinton broth (CA-MHB), and brain heart infusion broth (BHI) were purchased from Becton Dickinson (Franklin Lakes, NJ). Tryptic soy agar (TSA) plates, TSA plates with 5% sheep blood, and CA-MHB with 3% laked horse blood were purchased from Remel (San Diego, CA). MHB was used to assay all Staphylococcus aureus strains. CA-MHB was used for Staphylococcus epidermidis, Enterococcus faecalis, Enterococcus faecium, Escherichia coli, and Pseudomonas aeruginosa. CA-MHB supplemented with 3% laked horse blood was used for Streptococcus pneumoniae and Streptococcus pyogenes. Staphylococcus, Enterococcus, Escherichia, and Pseudomonas strains were cultured at 37°C with ambient air. Streptococcus strains were cultured at 37°C with 5% CO2 (24, 33). The separate conditions used for radiolabeled precursor uptake assays are described below.
Compounds, reagents, and radiochemicals.The methods used for the synthesis of compounds 1312 (10), 1575 (14), 1614 (14), and 1717 (14) have been previously described. The methods used for the synthesis of compounds 1962, 2062, 2093, 2114, and 2144 are described in the supplemental material. The following antibiotics were purchased commercially: vancomycin (Sigma-Aldrich, St. Louis, MO), linezolid (Chem-Impex International, Wood Dale, IL), rifampin (Chem-Impex International, Wood Dale, IL), ciprofloxacin (Acros Organics, Geel, Belgium), and novobiocin (Promega, Madison, WI). Ketoprofen was purchased from Sigma-Aldrich (St. Louis, MO). Human pooled serum was purchased from Thermo Fisher Scientific (Waltham, MA). Dulbecco's phosphate-buffered saline with calcium and magnesium (dPBS) was purchased from Lonza (Basel, Switzerland). [methyl-3H]thymidine (in 2% ethanol [EtOH], 69.7 Ci/mmol) and [5,6-3H]uridine (in sterile water, 60 Ci/mmol) were purchased from American Radiolabeled Chemicals (St. Louis, MO). l-[4,5-3H(N)]lysine (in 2% EtOH, 82.4 Ci/mmol) was purchased from PerkinElmer (Waltham, MA). Values of the logarithm of the partition coefficient (logP) were calculated with ChemAxon software (Cambridge, MA).
Production of recombinant S. aureus MetRSs.The SaMetRS gene (UniProtKB accession number A0A0H2XID2 ) was PCR amplified (sense primer, 5′-GGGTCCTGGTTCGGCTAAAGAAACATTCTATATAACAACCCCAATATAC-3′; antisense primer 5′-CTTGTTCGTGCTGTTTATTATTTAATCACTGCACCATTTGGAATTG-3′) from genomic DNA isolated from S. aureus (ATCC 29213) cultures. The PCR product was then cloned into the AVA0421 plasmid, and the sequence was verified. The expression of recombinant protein was performed as previously described (34). The N-terminal 6His fusion proteins were purified by nickel affinity chromatography followed by size exclusion gel chromatography (Superdex 75 26/60; GE Biosciences, Piscataway, NJ).
Enzyme assays.Inhibition of SaMetRS was measured using an ATP depletion assay as previously described (34) with some modifications. The compounds were preincubated for 15 min at room temperature in a 96-well plate with 400 μg/ml bulk E. coli tRNA, 25 nM SaMetRS, 0.1 U/ml pyrophosphatase, 0.2 mM spermine, 0.1 mg/ml bovine serum albumin, 2.5 mM dithiothreitol, 25 mM HEPES-KOH, pH 7.9, 10 mM MgCl2, 50 mM KCl, and 2% dimethyl sulfoxide (DMSO). The reagents were purchased from Sigma-Aldrich or Roche. The reaction was started with the addition of 150 nM ATP and 20 μM l-methionine, and after 120 min, incubation was stopped by the addition of an equal volume (50 μl) of the Kinase-Glo reagent (Promega). Percent inhibition was calculated as follows: 100 × [(RLU of test compound – RLU of positive control)/(RLU of negative control – RLU of positive control)], where RLU (relative light units) is measured on a Perkin Elmer Microbeta 2 scintillation counter. The positive control contains all assay reagents except compound. The negative control contains all assay reagents except compound and l-methionine. IC50s were calculated by nonlinear regression with a sigmoidal dose-response in Prism (version 3.0) software.
Bacterial strains.Strains with ATCC designations either were obtained directly from the American Type Culture Collection (Manassas, VA) or were kindly provided by the University of Washington Clinical Microbiology laboratory. Escherichia coli permeability mutants (the properties of which are shown in Table S1 in the supplemental material) were provided as a gift from Katherine Young at Merck (Rahway, NJ).
Macromolecular synthesis assays.The methods for measuring the uptake of radiolabeled precursors by S. aureus (ATCC 29213) were adapted from previous publications (21, 35, 36). For these assays, bacteria were grown in defined medium (DM), consisting of RPMI 1640, pH 7.3 ± 0.1, without phenol red or l-glutamine (Lonza, Basel, Switzerland) supplemented with 4 mM l-glutamine (Lonza, Basel, Switzerland), 10 mM HEPES (Lonza, Basel, Switzerland), and 1% (wt/vol) d-glucose (Sigma-Aldrich, St. Louis, MO). Fresh overnight cultures grown in DM at 37°C were diluted 1:50 in prewarmed DM and grown at 37°C with shaking (150 rpm) until they reached an optical density at 600 nm (OD600) of 0.420, correlating to ∼1 × 109 CFU/ml in mid-log phase. Each compound was assayed in quadruplicate with an 11-point 3-fold serial dilution per radioisotope. A prewarmed 96-well V-bottom plate (Corning 3894; Corning, Corning, MA) containing 25 μl of a 4× final concentration of test compound was inoculated with 65 μl of mid-log-phase bacteria (OD600, 0.420). Both positive- and negative-control wells received 25 μl of untreated DM and 65 μl of inoculum at the same time. After 1 min, 10 μl of radiolabeled precursor (10× final concentration in DM) was added to the samples and the positive-control wells. The final isotope concentrations for the assay of [3H]lysine (protein), [3H]thymidine (DNA), and [3H]uridine (RNA) were 10 μCi/ml, 2 μCi/ml, and 2μCi/ml, respectively. The plates were incubated at 37°C for 25 min and terminated by the addition of 50 μl of 30% trichloroacetic acid (TCA)–70% ethanol to all test and control wells. After termination, 10 μl of 10× radiolabeled precursor was added to the negative-control wells. The negative control consisted of radiolabeled precursors added after termination of the bacterial incubation in order to represent the background measurement of the isotope. The plates were sealed with plate tape (Thermo Fisher Scientific, Waltham, MA) and shaken at 250 rpm for 1 h at room temperature. Aliquots of 125 μl were transferred from the 96-well V-bottom plates to 96-well filter plates (Merck Millipore, Billerica, MA). To bind macromolecules, the samples were passed through a filter membrane (a 0.45-μm-pore-size hydrophilic Durapore polyvinylidene difluoride membrane) with a vacuum manifold, and then the filter was washed four times with 200 μl 10% TCA and once with 150 μl of 95% ethanol and dried overnight in a vacuum at room temperature. Twenty-five microliters of Ultima Gold scintillation fluid (PerkinElmer, Waltham, MA) was added to each well, and the disintegrations per minute were quantified using a MicroBeta2-2450 scintillation counter (PerkinElmer, Waltham, MA). The percent incorporation was determined by subtracting the number of disintegrations per minute for each well from the average number of disintegrations per minute for the negative background and dividing by the average positive disintegrations per minute incorporated and multiplying by 100. Error bars represent standard errors of the means (SEMs) between replicates. The assay was run twice, with similar results being obtained each time.
Susceptibility testing.MIC determinations were performed in triplicate in 96-well round-bottom microtiter plates (Corning, Corning, NY), as described by the Clinical and Laboratory Standards Institute (CLSI) (24, 33). Serial 2-fold dilutions of compounds were added to the plates in 50-μl volumes. An additional 50 μl of medium containing bacterial cells (1 × 106 CFU/ml) was then added to each well. The maximum DMSO concentration was 0.5%. The plates were incubated at 37°C for at least 18 h before the susceptibility result was read by determination of the absorbance (OD600) using a BioTek ELx800 absorbance microplate reader. The lowest concentration causing a ≥90% inhibition of growth compared to the growth for the untreated control was recorded as the MIC (and also corresponded to the visual MIC). MICs were measured at least twice, and the higher value (if the values were different) was recorded herein.
MBCs (defined as the concentration killing 99.9% of the inoculum) were determined according to published methods (36, 37). Using glass tubes (16 by 100 mm), serial 2-fold dilutions of compound were generated from DMSO stocks in single 1-ml volumes. Maximum DMSO concentrations were 0.5%. An additional 1 ml of medium with 1 × 106 CFU/ml was added per sample. Each experiment's inoculum was serially diluted and plated on TSA to count competent cells. Cultures were incubated at 37°C for at least 20 h and plated on TSA for determination of the number of CFU. Additionally, wells with each dilution of compound were sampled after 20 h of incubation for MIC determination, as described above.
Cytotoxicity testing on mammalian cells.Human cell lines CRL8155 (lymphoblasts) and HepG2 (hepatocellular carcinoma cells) were acquired from ATCC. The cultures were grown in RPMI 1640 medium with 10% fetal bovine serum, penicillin, and streptomycin at 37°C with 5% CO2. In 96-well plates, the cells were exposed to serial dilutions of the compounds for 48 h, and toxicity was quantified using alamarBlue (Thermo Fisher Scientific, Waltham, MA) (37). Assays were performed in quadruplicate, and EC50s were calculated by nonlinear regression methods using the software of the Collaborative Drug Database (Burlingame, CA).
Determination of resistance rates.The spontaneous rates of resistance to the test compounds were determined according to published methods (35, 38). Agar selection plates were made by adding compound from DMSO stocks into molten Mueller-Hinton agar in a 55°C water bath. Each compound used four plates (150 by 15 mm; catalog number P5981-100EA; Sigma-Aldrich, St. Louis, MO) containing 4× or 8× the MIC of the compound. The final DMSO concentration was <0.1% per plate. Plates were dried in a sterile hood for 30 min prior to overnight storage at 4°C and prewarmed in a 37°C incubator for 1 h prior to assay.
A fresh overnight culture was diluted 1:50 in MHB and grown at 37°C with shaking (150 rpm) until it reached an OD600 of 0.4, correlating to ∼2 × 109 CFU/ml. Approximately 3 ml, for a total of 6 × 109 CFU, was distributed onto 4 plates for each compound. The plates were incubated at 37°C for 72 h prior to counting of the colonies. The starting inoculum was also serially diluted and plated to quantify the initial bacterial load. The resistance frequency was determined as the number of compound-resistant colonies divided by the total number of colonies plated.
Serum shift assays.To assess the role of protein binding on compound susceptibility, MIC determinations were performed in triplicate in the presence and absence of 50% human serum (39, 40). Serial 2-fold dilutions of a 2× concentration of compound were generated in MHB, and 50 μl was aliquoted onto 96-well plates with a DMSO concentration limit of 0.5%. The bacteria were adjusted to 1 × 108 CFU/ml in MHB and then separately further diluted 1:100 in MHB and 100% heat-inactivated filter-sterilized pooled human serum. Fifty microliters of the bacterial suspension was added to each well of the corresponding plates, and the plates were incubated at 37°C for ∼20 h. The lowest concentration causing ≥90% growth inhibition was recorded as the MIC. The median value from three independent assays is reported.
Protein binding assays.Compound binding to mouse plasma proteins was determined using 96-well equilibrium dialyzer plates (catalog number SDIS 9610EN; Nest Group, Inc.). Mouse plasma (catalog number MSEPLLIHP-SW-F; BioreclamationIVT, Westbury, NY) containing compound (final concentration, 1 μM) was added to a donor chamber as a 150-μl volume. The buffer solution (0.2 mM phosphate buffer, 150 μl) was added to the reciprocal acceptor chamber. Each compound was tested in triplicate. To prepare calibration solution for compound quantifications, blank wells containing only mouse plasma in a donor well and buffer solution in its acceptor well were prepared. The equilibrium dialysis was carried out by rocking the plate for 22 h at 37°C. Once equilibrium was reached, the plasma and buffer solutions from both wells were carefully removed for further analysis by liquid chromatography-tandem mass spectrometry. Plasma solution and the internal standard were mixed in the presence of 80% acetonitrile. After centrifugation of the solution, the supernatant was transferred to a liquid chromatography insert. Similarly, the buffer solution from the acceptor side was prepared in the presence of 40% acetonitrile. Calibration standards for the donor and acceptor sides were prepared with compound concentrations of 50 nM, 100 nM, 250 nM, 500 nM, and 1 μM. The compound concentrations from each well were calculated from the calibration curves using Microsoft Excel software. The percentage of the test compound bound was determined as follows: Percent free compound = (concentration in the buffer chamber/concentration in the plasma chamber) × 100 and Percent bound compound = 100 − percent free compound.
Metabolic stability.Liver microsome stability assays were done by a contract research laboratory, Wuxi AppTec Co. (Hubei, China). Briefly, the compounds at a 1 μM concentration were incubated singly with human or CD-1 mouse liver microsomes for 6 times (0, 5, 10, 20, 30, and 60 min). Loss of the parent compound was quantified by liquid chromatography-tandem mass spectrometry. The measured half-lives of the control compounds (testosterone, diclofenac, and propafenone) were within the expected ranges.
Murine pharmacokinetics studies.Nonfasted female Swiss Webster mice (age, 6 to 8 weeks) in groups of three mice each were administered compounds by oral gavage (10 mg/kg p.o.) or by retro-orbital injection (5 mg/kg i.v.). MetRS inhibitors were dosed in vehicle consisting of 7% Tween 80, 3% ethanol, and 0.9% saline. Linezolid was dosed in a vehicle consisting of 0.5% methylcellulose (catalog number cP400; Sigma-Aldrich, St. Louis, MO), 0.5% Tween 80, and 0.9% saline. The times of blood collection were as follows: for oral administration, 30, 60, 120, 240, 360, 480, and 1,440 min, and for i.v. administration, 5, 15, 30, 60, 240, 360, 480, and 1,440 min. Tail blood was collected in heparinized capillary tubes, and 20 μl was spotted onto Whatman FTA DMPK-C cards (GE, Fairfield, CT). The whole-blood samples were extracted with acetonitrile for measurement of the compound concentrations by liquid chromatography-tandem mass spectrometry. The values of the pharmacokinetic parameters were calculated using Phoenix WinNonlin (version 6.3) software (Certara, Princeton, NJ).
Murine thigh infection model.Animal studies were approved by the Institutional Animal Care and Use Committee at the University of Washington, Seattle, WA. The methods were based on those published in the literature (41–44). Female specific-pathogen-free CD1 mice were obtained from Charles River (Wilmington, MA) and weighed 23 to 27 g. They were allowed at least 3 days to acclimate prior to the study. Mice had access to food and water ad libitum. Neutropenia was induced by administering cyclophosphamide monohydrate (catalog number C7397; Sigma-Aldrich, St. Louis, MO) via intraperitoneal injection 4 days (at 150 mg/kg) and 1 day (at 100 mg/kg) prior to infection. Neutropenic status was confirmed by a neutrophil count of <100 cells/mm3. An overnight culture of S. aureus (ATCC strain 29213) was diluted 1:100 in MHB and incubated until it reached mid-log phase (OD600 < 0.750). The inoculum was prepared by pelleting of the log-phase culture and resuspension in sterile dPBS. The culture was adjusted to an OD600 of 0.200 and diluted 1:100 in sterile dPBS, correlating to an inoculum of ∼5 × 105 CFU/50 μl. The mice were infected by intramuscular injection of 50 μl in the right posterior thigh while they were under isoflurane gas anesthesia. At 1 h postinfection, one vehicle group was sacrificed for determination of the initial inoculum (stasis level of infection). Mice were dosed with the test compounds at 2 and 12 h postinfection, as follows. Vancomycin was given at 100 mg/kg subcutaneously in 100 μl in a 0.9% saline solution. Linezolid was administered at 75 mg/kg p.o. in 200 μl of 0.5% methylcellulose (catalog number cP400; Sigma-Aldrich, St. Louis, MO), 0.5% Tween 80 (Sigma-Aldrich, St. Louis, MO) in distilled water (45). MetRS inhibitors were administered at 75 mg/kg p.o. in 200 μl of vehicle containing 60% phosphatidylcholine solubilized in a carrier system (Phosal 53 MCT; Lipoid, Ludwigshafen Germany), 30% polyethylene glycol 400 (Sigma-Aldrich, St. Louis, MO), and 10% EtOH. Mice were sacrificed at 24 h postinfection; the thigh muscle was sterilely removed, weighed, homogenized in 5 ml dPBS, serially diluted, plated on tryptic soy agar in duplicate, and incubated overnight at 37°C. Colonies were counted to quantify the bacterial load as the number of CFU per gram of thigh tissue.
Sequence alignments.Global pairwise amino acid sequence alignments were generated with the NCBI alignment tool Clustal Omega (46).
ACKNOWLEDGMENTS
The research was supported by a grant from the CoMotion Innovation Fund at the University of Washington and by the National Institutes of Health (grant AI097177).
We acknowledge Jennifer McCullar and Lynn Silver for consulting on the project direction and experimental design. We thank Nicole Duster and Stephen Nakazawa Hewitt for cloning and purifying the SaMetRS and Uyen Nguyen for technical assistance with the microbiology work.
FOOTNOTES
- Received 12 May 2017.
- Returned for modification 3 June 2017.
- Accepted 22 August 2017.
- Accepted manuscript posted online 28 August 2017.
Supplemental material for this article may be found at https://doi.org/10.1128/AAC.00999-17 .
- Copyright © 2017 American Society for Microbiology.
