Peptide deformylase (PDF), a highly conserved metalloproteinase, has been observed to be critical to the maturation of proteins during translation in prokaryotic cells and was first described in Escherichia coli and Bacillus subtilis (1, 22). As formyl-methionyl-tRNA initiates protein synthesis in bacterial cells, the forming polypeptide chain characteristically contains an N-formyl-methionine residue at the N terminus. As the peptide chain elongates, the PDF enzyme removes the formyl group, permitting protein maturation through the subsequent removal of the methionine residue by a second enzyme, methionine aminopeptidase. Failure to remove the N-formyl group prevents the action of methionine aminopeptidase, which is required for cell growth. The PDF enzyme can also be present in eukaryotic cells, specifically within mitochondria, but remains much less active than in bacteria, thus making it an attractive target for antibacterial agents (32, 34).

Inhibitors of PDF have been described in recent years, with the naturally occurring antibacterial agent actinonin being a typical example (7, 13). LBM415 (also known as NVP PDF-713) is the first of the PDF inhibitor class to advance to clinical trials for the oral and parenteral treatment of respiratory tract and skin and skin structure infections caused by susceptible gram-positive and -negative organisms (6). The compound is an N-alkyl urea hydroxamic acid with the chemical name (S)-1-{(R)-2-[formyl-hydroxy-amino)-methyl]-hexanoyl}-pyrroli-dine-2-carboxylic acid (5-fluoro-1-oxy-pyridin-2-yl)-amide (Fig. 1). Recent spectrum-of-activity evaluations of the targeted species, including resistant subsets, have demonstrated a potential role for LBM415 in the treatment of the indicated infections (6, 9, 11, 19, 20, 33).

FIG. 1. Chemical structure of LBM415, a novel PDF inhibitor.

In this report we summarize the results of testing LBM415 and selected comparator agents against a worldwide collection of contemporary, clinical isolates chosen to overrepresent current resistance trends. This collection of 1,306 isolates was tested by standardized reference methodologies (agar and broth microdilution), and categorical interpretations were made using current National Committee for Clinical Laboratory Standards criteria (26). Additional LBM415 microbiologic features examined included the determination of bactericidal concentrations, rate of bacterial killing, and interactions (synergy testing) with other antimicrobials. Lastly, both single- and multiple-step mutational rates were determined to assess the ability of various bacterial species to develop resistance to LBM415.

Bacterial strains. The organisms (1,306 strains) selected to challenge the activity and spectrum of LBM415 included 153 staphylococci (90 oxacillin resistant [OR]), 170 S. pneumoniae strains (65, 52, and 53 penicillin susceptible, intermediate and resistant, respectively), 69 β-hemolytic streptococci strains (42 erythromycin resistant), 81 viridans group Streptococcus spp. strains (only 40.7% susceptible to penicillin), 74 vancomycin-susceptible enterococci, 30 vancomycin-resistant enterococci (only 53.3% susceptible to quinupristin-dalfopristin), 300 H. influenzae strains (130 ampicillin resistant), 103 Moraxella catarrhalis strains (90.3% penicillin resistant), 112 Enterobacteriaceae strains (14 species or genus groups), 107 nonfermentative gram-negative bacilli (six species or genus groups), 31 anaerobes, 50 Legionella pneumophila strains, and 26 other gram-positive cocci (see Table 3, below, for details). All isolates were derived from clinical specimens collected in 2001-2002 and identified by the submitting laboratories, with identifications confirmed using standard biochemical algorithms, including use of the Vitek System (bioMerieux, Hazelwood, Mo.).

TABLE 3. Summary of antimicrobial activities at individual MIC values of LBM415 tested against 11 organism groups

Antimicrobial agents. Compound LBM415 was obtained from Novartis Pharmaceuticals (Summit, N.J.). Comparison agents were provided by the manufacturers in the United States or by Sigma Chemical (St. Louis, Mo.). Comparison agents included amoxicillin-clavulanate, ampicillin, azithromycin, ceftriaxone, chloramphenicol, ciprofloxacin, clarithromycin, clindamycin, doxycycline, erythromycin, gentamicin (high-level screen, ≤500 μg/ml), levofloxacin, linezolid, penicillin, quinupristin-dalfopristin, rifampin, streptomycin (high-level screen, ≤1,000 μg/ml), teicoplanin, tetracycline, and vancomycin. Each antimicrobial agent was diluted in the appropriate medium (Mueller-Hinton [MH] broth with or without 5% lysed horse blood, or haemophilus test medium) and dispensed in reference microdilution trays (25).

Antimicrobial susceptibility testing. All tests were performed by a reference broth microdilution method (25) except for anaerobes and L. pneumophila, for which the agar dilution test was used (27). Brucella blood agar was used for testing anaerobes, and buffered yeast extract (BYE) agar with and without charcoal was used for testing of legionellae.

Minimum bactericidal concentrations were determined by methods described previously (10, 28). Kill-curve experiments used initial inoculum densities of 5 × 10⁵ CFU/ml and drug concentrations at two, four, and eight times the MIC. Timed samples were taken at baseline and 1, 2, 4, 8, and 24 h. Drug interaction (synergy) tests used codrugs (gentamicin, rifampin, ampicillin, and vancomycin) at concentrations of one-fourth their respective MIC combined with LBM415 at two, four, and eight times the MIC. Antimicrobials were defined as bactericidal if the initial inoculum was reduced by ≥3 log₁₀ CFU/ml within the monitored interval (24 h). The bacteriostatic level was a 0 to <3 log₁₀ CFU/ml change in the original inoculum (10, 16). Twenty organisms were monitored for bactericidal action by LBM415 tested alone, and 14 were tested in various combinations (21). When using a codrug combined with LBM415 for the synergy analyses, synergy was defined as a ≥2 log₁₀ CFU/ml decrease in the inoculum at any monitored time compared to the activity of LBM415 alone (four times the MIC). Antagonistic interactions were a ≥2-log₁₀ CFU/ml increase in growth for the combination compared to the colony count with LBM415 alone.

MIC methods for H. influenzae were tested in haemophilus test medium broth for LBM415 alone and with two efflux inhibitors (reserpine and phe-arg-β-napththylamide [MC207,110]; Sigma Chemical Co.) at concentrations of 10 and 20 μg/ml (29). Five strains were compared against both inhibitors, and 20 additional strains were screened with LBM415 combined only with phe-arg-β-naphthylamide.

The effects of changing standardized susceptibility testing conditions on the LBM415 MIC results were assessed by reference agar dilution methods and compared to broth microdilution test results (25). The MIC tests were performed using four different inoculum concentrations (10³, 10⁴, 10⁵, and 10⁶ CFU/spot), three incubation environments (anaerobic, ambient air, and 5% CO₂), three media (5% sheep blood, chocolated blood agar, and unsupplemented MH agar), three medium pHs (6.0, 7.2 to 7.4, and 8.0), and two concentrations of the divalent cation calcium (25 and 50 μg/ml). The MH agar test results conforming to the NCCLS standard were compared to MICs obtained in MH broth by the reference microdilution method (25).

Spontaneous mutation rates and passaging studies. Single-step mutational rates to resistance were determined by plating a 10⁸ CFU organism suspension (20 strains) onto MH agar plates containing LBM415 at concentrations two, four, and eight times the established MIC (25). The frequency of mutation was determined by colony counts at 24 and 48 h.

Multiple-step mutational rates were established via daily passaging of strains in subinhibitory concentrations of LBM415 in broth microdilution panels. After determination of baseline MICs, inocula for the subsequent MIC test were taken from wells 1 log₂ dilution below the MIC. Passages were continued for 10 consecutive days using 10 strains from eight different species groups.

LBM415 activity against gram-positive species. Table 1 presents results of testing LBM415 and eight other agents against staphylococci (153 strains), streptococci (320 strains), and enterococci (130 strains). All LBM415 MICs were ≤2 μg/ml for S. aureus, with MICs at which 50% and 90% of isolates were inhibited (MIC₅₀ and MIC₉₀ values) of 1 and 2 μg/ml and 0.5 and 2 μg/ml for oxacillin-susceptible (OS) and OR strains, respectively. All comparator drugs were effective in vitro against OS S. aureus (OSSA), with susceptibilities ranging from 83% for erythromycin to 100% for the five other agents. For the 51 strains of ORSA, however, only LBM415, vancomycin, teicoplanin, and linezolid (MIC₉₀ results, 2 μg/ml; 100% susceptible) and quinupristin-dalfopristin (MIC₉₀, 0.5 μg/ml; 98% susceptible) remained highly active.

TABLE 1. Comparative activity screen of LBM415, a deformylase inhibitor, and eight selected comparison classes of antimicrobial agents tested against staphylococci (153 strains), streptococci (320 strains), and enterococci (130 strains)

CoNS strains were slightly less susceptible to LBM415 than were S. aureus strains, with the MIC ranges extending to 2 μg/ml for OS strains and to 4 μg/ml for OR strains. All comparator agents inhibited greater than 90% of OS strains at susceptible breakpoints (exception for levofloxacin, 80% susceptible), whereas among OR strains only doxycycline (87.2% susceptible) and the antimicrobial group consisting of linezolid, quinupristin-dalfopristin, and the glycopeptides (all 100% susceptible) remained highly active.

Among all S. pneumoniae strains, regardless of penicillin susceptibility, LBM415 was uniformly active, with MIC₅₀s of 0.25 to 0.5 μg/ml, MIC₉₀s of 0.5 to 1 μg/ml, and a MIC range from ≤0.016 to 2 μg/ml. All strains were uniformly susceptible to levofloxacin, vancomycin, and quinupristin-dalfopristin (100%) and increased resistance documented to ceftriaxone, erythromycin, clindamycin, and chloramphenicol among penicillin-intermediate and -resistant strains.

LBM415 was also uniformly active against β-hemolytic and viridans group streptococci (MIC₉₀s, 1 and 0.5 μg/ml). While the β-hemolytic streptococci were 100% susceptible to the other comparators (with the exceptions of erythromycin and clindamycin, 85.5 and 94.2% susceptible, respectively), the viridans group strains displayed variable resistance to all comparators except for linezolid and vancomycin (100% susceptible).

No difference was noted in the activity of LBM415 against vancomycin-susceptible or -resistant enterococci (Table 1). Among vancomycin-susceptible enterococci, only teicoplanin and linezolid were more potent (MIC₉₀, ≤4 μg/ml) than LBM415; among vancomycin-resistant strains, only linezolid remained more potent (MIC₉₀, 2 μg/ml).

LBM415 displayed a broader range of activity (0.03 to >32 μg/ml) against a collection of other gram-positive species, including Aerococcus, Bacillus, Corynebacterium, Gemella, Lactobacillus, Lactococcus, Leuconostoc, Listeria, Micrococcus, Nocardia, and Stomatococcus, although the group MIC₅₀ and MIC₉₀ values (0.25 and 8 μg/ml, respectively) remained within clinically achievable concentrations.

LBM415 activity against gram-negative species. LBM415 displayed activity against both H. influenzae (300 strains) and M. catarrhalis (103 strains) (Table 2), although the MIC₅₀ and MIC₉₀ of both ampicillin-susceptible and ampicillin-resistant H. influenzae (1 and 4 μg/ml and 2 and 8 μg/ml, respectively) were 8- to 16-fold higher than those for M. catarrhalis (0.25 and 0.5 μg/ml, respectively). A slight trend in decreased activity among ampicillin-resistant H. influenzae strains was apparent, with a twofold decrease in potency over ampicillin-susceptible strains being documented. All H. influenzae and M. catarrhalis strains remained highly susceptible to the comparator agents used clinically.

TABLE 2. Comparative antimicrobial activity screen of LBM415, a deformylase inhibitor, tested against H. influenzae (300 strains), M. catarrhalis (103 strains), and L. pneumophilia (50 strains)

A total of 50 strains of L. pneumophila were tested against LBM415 and three selected comparison agents often used for legionellosis (Table 2). The activity of the PDF inhibitor was equal to that of levofloxacin (MIC₉₀, 0.12 μg/ml) and eightfold more potent than erythromycin. The use of charcoal in the BYE test medium markedly reduced LBM415 activity.

Summary of antimicrobial activity of LBM415. The cumulative percentages of bacterial strains inhibited at specific MICs are displayed in Table 3. Among the most common bacterial pathogens of community-acquired pneumonia and skin and skin structure infections, all (100%) staphylococci, streptococci, enterococci, M. catarrhalis, and L. pneumophila strains were inhibited at 8 μg or less of LBM415/ml. H. influenzae strains were only slightly less susceptible (97%) to LBM415. Among other bacterial groups, 100% of gram-positive and -negative anaerobic strains were inhibited by ≤4 μg of LBM415/ml, whereas Enterobacteriaceae and most nonfermentative gram-negative bacilli were not inhibited by the PDF inhibitor compound at readily achievable concentrations.

Effects of changing in vitro test conditions on the LBM415 MIC. The effects on the LBM415 MIC of changing in vitro testing parameters were assessed using 10 gram-positive organisms (five species). Variations from the reference MIC result (on MH agar or in MH broth, pH 7.2 to 7.4, incubated in ambient air, calcium at 25 μg/ml, and an inoculum of 10⁴ CFU/spot) were observed as follows: (i) a trend toward a 1-log₂ dilution-higher MIC with the agar-based method; (ii) two- to fourfold MIC increases with elevated inoculum concentrations (10 to 100 times) and lower MIC results with a reduction in the inoculum to 10³ CFU/spot; and (iii) varying the incubation environment, medium supplement, pH, or calcium concentration had little influence on the LBM415 MIC result (data not shown).

Effects of efflux inhibitors on the MIC for H. influenzae. Due to the presence of an extended MIC distribution, LBM415 was tested against H. influenzae strains using media containing two efflux inhibitors that target different membrane pumps (29). Initial screening of two inhibitory concentrations of each drug (five strains) demonstrated no significant inhibition (a lowering of the LBM415 MIC by ≥4-fold) by reserpine. However, phe-arg-β-naphthylamide showed a significant pump inhibition at both tested concentrations against strains with higher LBM415 MICs. Follow-up tests with 20 H. influenzae strains having varied LBM415 MICs showed an average of 1-log₂ dilution decrease in the MIC when combined with ≥10 μg of phe-arg-β-naphthylamide/ml. Efflux pumps appear to influence the level of H. influenzae susceptibility to PDF inhibitors.

Spontaneous mutation rates and passaging experiments. The results of the single-step mutation tests at eight times the MIC of LBM415 are listed in Table 4. Generally, the rates were extremely low at <1.5 × 10⁻⁸ for enterococci, Streptococcus pyogenes, S. pneumoniae, M. catarrhalis, H. influenzae, and some strains of S. aureus. Higher mutational frequencies (3.1 × 10⁻⁶ to 3.0 × 10⁻⁷) were noted for single strains of OSSA or ORSA and CoNS. One strain of ampicillin-susceptible Enterococcus faecalis had a single-step mutational rate of 2.2 × 10⁻⁶, yet another tested vancomycin-resistant strain had a rate of <1.5 × 10⁻⁸.

TABLE 4. Single-step mutational rates at eight times the MIC of LBM415 for 20 strains of species inhibited by this new compound

Ten organisms were passaged in subinhibitory concentrations of LBM415 for 10 days. Only three strains had a >4-fold increase in the LBM415 MIC over the study interval (E. faecium and CoNS, 0.5 to 4 μg/ml; an S. aureus strain, 0.5 to >32 μg/ml). The strain of OSSA demonstrating the greatest MIC increase also had the highest single-step mutational rate (3.0 × 10⁻⁷) (Table 4). All other species had LBM415 MIC results that were unchanged or no greater than a fourfold increase (average, twofold MIC elevation).

Bactericidal activity and synergy studies. Twenty organisms from 11 species or genus groups were tested by kill-curve methods against three concentrations of LBM415 (two, four, and eight times the MIC). Table 5 summarizes the results, which demonstrated a dominant bacteriostatic action for this new PDF inhibitor. Bactericidal action at 24 h, usually at concentrations four or eight times the MIC, was observed with four strains (20%; ORSA, OS CoNS, penicillin-resistant S. pneumoniae, and H. influenzae).

TABLE 5. Time-kill analysis of 20 organisms tested against LBM415 at three concentrations (two, four, and eight times the MIC)

Drug interaction (synergy) results were derived from 28 kill-curve tests for 14 organisms. Only two studies showed synergy (LBM415 plus gentamicin versus S. mitis; LBM415 plus ampicillin versus vancomycin-resistant E. faecalis). All other interactions were classified as indifferent, and no antagonism was observed.

While a number of antimicrobial classes (tetracyclines, aminoglycosides, macrolides, lincosamides, streptogramins, and oxazolidinones) target translational bacterial metabolic pathways, PDF inhibitors constitute a novel class of antimicrobial agents by targeting the enzyme responsible for removal of the N-terminal formyl group from the majority of bacterial proteins. Failure to deformylate nascent proteins results in the inability of other enzymes to complete the protein maturation process, resulting in bacteriostasis (4). The antibacterial activity of actinonin has been known for over 40 years, and only recently has the actual mechanism of action as an inhibitor of PDF been more fully elucidated (7, 14). Actinonin and its derivatives have not been developed for human clinical use because of their poor pharmacokinetic properties (8).

More recently, screening of large compound collections and fabrication of new agents have revealed the presence of a number of candidate molecules that function as PDF inhibitors, especially the N-alkyl urea hydroxamic acids, some of which have pharmacokinetic characteristics more amenable to clinical applications (4, 8, 15, 31). Many of these have shown significant activity against respiratory tract pathogens, including S. pneumoniae and M. catarrhalis, with variable activity against H. influenzae and Chlamydia spp. (4, 6, 30, 33). The PDF inhibitor NVP-PDF386 was recently tested against more than 1,000 clinical isolates for potency and spectrum of activity. MIC₅₀ and MIC₉₀ results (in micrograms per milliliter) for NVP PDF-386 were as follows: S. aureus, 0.5 and 1; CoNS, 0.5 and 1; S. pneumoniae, 0.25 and 0.5; other streptococci, 0.25 and 0.5; Enterococcus spp., 1 and 2; M. catarrhalis, 0.25 and 0.25; H. influenzae, 8 and 32; no activity against Enterobacteriaceae and nonfermentative gram-negative bacilli (>32 and >32 for both groups) (18). No differences were noted between subsets of staphylococci susceptible and resistant to oxacillin, nor to pneumococci susceptible and resistant to penicillin.

Likewise, LBM415 is a promising PDF inhibitor in an oral formulation being proposed for the treatment of community-acquired respiratory tract infections and various uncomplicated and complicated gram-positive infections. Previous studies have documented the in vitro utility of LBM415 with a variety of gram-positive clinical isolates, including strains resistant to linezolid or quinupristin-dalfopristin (19, 20). The spectrum of activity in the present study appears particularly suited for these indications, with 100% inhibition (≤8 μg/ml) of staphylococci, streptococci, enterococci, M. catarrhalis, and L. pneumophila strains along with 97% of H. influenzae strains. Oxacillin resistance among staphylococci, penicillin resistance among S. pneumoniae strains, and vancomycin resistance among enterococci had no effect on LBM415 potency. Ampicillin-resistant H. influenzae strains did display a modest twofold increase in MICs compared to ampicillin-susceptible strains.

Of the targeted bacterial species tested, H. influenzae consistently displayed the highest MIC results but remained at a level (MIC₉₀, 4 μg/ml) where most strains would be expected to respond given favorable pharmacokinetic-pharmacodynamic studies, as reported for another related PDF inhibitor (W. A. Craig and D. Andes, Abstr. 41st Intersci. Conf. Antimicrob. Agents Chemother., abstr. F-355, 2002). Studies using the efflux inhibitor phe-arg-β-naphthylamide were able to demonstrate that decreased activity of LBM415 with this species was influenced by efflux pumps, demonstrating the contribution of this common mechanism to the elevated MICs and potential resistance.

As with virtually all antimicrobial agents, the development of resistance can be expected to occur. The spontaneous mutation rates found in the present study generally occurred at extremely low levels (≤1.5 × 10⁻⁸) for targeted species, although other single staphylococcal strains did demonstrate higher mutational frequencies (Table 4). With the exception of one strain of S. aureus, all species and strains that were passaged in subinhibitory concentrations of LBM415 for 10 days had MIC results that were unchanged or averaged only a twofold elevation, suggesting that exposure to the compound during a therapeutic trial will be unlikely to result in the rapid emergence of resistance.

Studies with the PDF inhibitor BB-3497 demonstrated that resistance developed through mutations within the bacterial formyltransferase gene (fmt), as had been shown to occur previously with actinonin, resulting in a lack of transformylation activity (8, 24). E. coli and S. aureus resistant mutants have readily been isolated in vitro following growth on BB-3497-containing plates at two and four times the MIC for each and arose at frequencies of 1 × 10⁻⁷ and 2 × 10⁻⁷, respectively. In all cases, mutations within the fmt gene resulted in the expression of a truncated and presumably nonfunctional formyltransferase (8). Actinonin-resistant S. aureus strains were observed to grow slowly in vitro and produced attenuated infection in a murine abscess model, requiring 25- to 100-fold-higher inoculation amounts (24). The frequency of PDF resistance of S. pneumoniae to actinonin was found to be 10⁻⁸, with the mechanism of resistance resulting in a modification of the enzyme target rather than in the lack of transformylation (23).

Previous studies have documented that PDF inhibitors are bacteriostatic (4, 8). Time-kill analysis as performed here confirms that assessment, with a dominant bacteriostatic action being evident (Table 5). Bactericidal activity, when present, occurred only when the organisms were exposed to four and eight times the LBM415 MIC and was observed in only 20% of strains tested. Drug interaction studies with ampicillin, gentamicin, rifampin, and vancomycin did not identify any class-specific synergistic interactions to enhance bactericidal action with any of the organisms tested. In only two instances was synergy apparent; all other interactions were considered as being indifferent. Importantly, no antagonism was observed.

PDF inhibitors are a novel class of antimicrobial with bacteriostatic properties and are rapidly being developed for use as a therapeutic alternative for agents used in the outpatient arena in the treatment of community-acquired respiratory tract infections and for infections of the skin and skin structures. In support of the accurate assessment of susceptibility testing profiles among patient bacterial isolates recovered during clinical trials, quality control guidelines for MIC and disk diffusion testing of LBM415 have recently been proposed (2, 3). Validation of commercial dry-form susceptibility testing panels containing LBM415 has also been performed, assuring the reliability of the product when testing clinical trial isolates (12).

The spectrum of activity and potency of LBM415 for use against the targeted bacterial species appear to be adequate, and the development of resistance can be expected to occur no more rapidly than for other indicated antimicrobials. The continued, accelerated development of PDF inhibitors promises to be a significant advance in therapeutics should the pharmacokinetic-pharmacodynamic and toxicity parameters in humans support use of the class for respiratory and skin and skin structure infections, among other potential indications.

We express our appreciation to the following individuals for assistance with technical support, manuscript preparation, and editorial processing: K. Meyer, P. Rhomberg, D. Biedenbach, and K. Bracken.

This study was supported by an educational/research grant from Novartis Pharmaceuticals, Inc.