Prior Antibacterial Peptide-Mediated Inhibition of Protein Folding in Bacteria Mutes Resistance Enzymes

Resistance to antibiotics is developing at an alarming rate (22), and this trend involves all antimicrobial drug families (15). While resistance to fluoroquinolones or aminoglycosides is a major public health concern (2, 8), the inactivity of β-lactam antibiotics poses an economic burden to low-cost prescription patterns (9). Resistance to β-lactams is primarily due to hydrolysis of the ring by β-lactamases (16), of which the TEM-1 enzyme is the most commonly encountered resistance marker in members of the family Enterobacteriaceae (24).

Designer analogs of the small, proline-rich antimicrobial peptide pyrrhocoricin kill bacteria that are resistant to β-lactams (5), aminoglycosides (6), or fluoroquinolones (21). The mode of action of these peptide derivatives is inhibition of the 70-kDa heat shock protein (Hsp) DnaK and, thus, protein folding (14) combined with membrane disintegration (20). The peptides fail to bind to mammalian Hsp70 and are nontoxic to mammalian cells or experimental animals (6, 13, 21). Due to the renal clearance of peptide drugs (17) and the low-salt environment of the kidneys (10), we initially concentrated on Escherichia coli and Klebsiella pneumoniae strains, bacteria dominant in urinary tract infections (UTIs) (26). Bacterial Hsps are targets of natural antibacterial products, including xylitol (12), and inhibition of heat-induced Hsp expression leads to cytotoxic effects to bacteria (1).

β-Lactamases can be inactivated by compounds that bind to their active site (4, 23). Proline-rich antibacterial peptides inactivate test bacterial enzymes (14), and our idea was to use them to deplete all therapeutically relevant enzyme populations. In the study described in this report, we studied how sequential or simultaneous addition of the most active peptide derivative, A3-APO, and amoxicillin to clinical E. coli cultures of UTI origin expressing the TEM-1 β-lactamase influences the MIC. The control antibiotic was the aminoglycoside kanamycin, the main mechanism of resistance to which proceeds through mutation in chromosomal genes (18). Fluoroquinolones target DNA gyrase and topoisomerase II, and the resistance pathway involves alterations in the target sequences (11). Yet, we also included ciprofloxacin because fluoroquinolones induce DnaK production in E. coli (19).

Overnight cultures of amoxicillin-resistant, pyrrhocoricin-sensitive E. coli strains SEQ102 and BF1023 (6), grown in tryptic soy broth (TSB), were diluted 10-fold in TSB-ampicillin and grown for 3 h at 37°C. The cells were harvested and lysed, and the clear lysates were loaded onto Ampholine PAG plates 3.5-9.5 (GE Healthcare). An isoelectric focusing gel was run at 1,000 V for 1.5 h, and β-lactamases were visualized by 0.5 mg/ml nitrocefin. Figure 1 shows that both E. coli test strains expressed high levels of the TEM-1 enzyme. The other two strains were E. coli S5081, which is resistant to kanamycin (6), and ciprofloxacin-resistant strain E. coli 045-849 from the SENTRY collection (21).

The MICs and 50% inhibitory concentrations (IC₅₀s) of A3-APO and the small-molecule antibiotics were determined by the broth microdilution liquid growth inhibition method (3). Growth inhibition assays were performed with sterile Nunc 96-well plates in a final volume of 100 μl (5). The cell concentrations were estimated from the absorbance at 600 nm with the formula CFU/ml = A₆₀₀(3.8 × 10⁸), where CFU is the number of CFU. Ninety microliters of mid-logarithmic-phase bacterial cultures (4 × 10⁵ CFU/ml) in Mueller-Hinton broth was added to 10 μl of serially diluted antibiotics (0.5 to 64 μg/ml) dissolved in sterilized water. The cultures were incubated at 37°C for 16 to 20 h, and growth inhibition was measured by determination of the absorbance at 600 nm. The MIC was identified as the lowest antimicrobial dose when the absorbance readings did not exceed the values obtained in medium only. The IC₅₀ data were calculated from the growth inhibition curves (Fig. 2) to identify sublethal doses of A3-APO for the subsequent synergy assay. Table 1 shows the sequence of the A3-APO peptide and the MIC and IC₅₀ efficacy data. A3-APO exhibited MICs between 32 and 64 μg/ml and IC₅₀ values between 4 and 8 μg/ml. Thus, in the synergy assay, the peptide derivative was added as low and high concentrations of 2 to 4 and 8 to 16 μg/ml, respectively. It is relevant here that the subcutaneous administration of 5 mg/kg pyrrhocoricin-based dimers to mice yields 21 μg/ml peptide in blood over a 2-h window, representing a therapeutic index of at least 100 (21). The conventional antibiotics were mostly inactive against their respective strains.

In the synergy assay, the bacteria were preincubated with the A3-APO peptide for 1 h to inhibit protein folding. After 1 h the antibiotics were added and growth inhibition was measured. Figure 2A documents that amoxicillin alone was inactive against E. coli BF1023 but that its activity could be fully recovered when the culture was preincubated with 16 μg/ml A3-APO. The synergy was not observed when the peptide and the small-molecule antibiotic were added simultaneously. In the latter case, the number of remaining bacterial colonies hovered around the value obtained with the antibacterial peptide alone (25 to 30%), indicating that β-lactamase activity was manifested before the enzyme's folding could be inhibited. Likewise, no synergistic effect was observed at the lower peptide concentration (approximately 40% killing). To compensate for the minor variations in the MIC determinations, all growth inhibition assays for which the results are shown in Fig. 2A to D were run on single plates. The assays were repeated twice, with nearly identical results each time, although all individual assays showed slight experimental errors at extreme antibiotic concentrations.

The results of the experiments with E. coli SEQ102 mirrored the results with the other β-lactam-resistant isolate (Fig. 2B). The antimicrobial activity of amoxicillin could be fully recovered in a peptide dose- and preincubation regimen-dependent manner with 8- and 4-μg/ml active and inactive peptide concentrations, respectively. The results were completely different with kanamycin and E. coli S5081 (Fig. 2C), with which no activity of the small-molecule antibiotic could be recovered under any circumstance. The bacterial growth data in the synergy lanes reflect the efficacy of A3-APO alone at both concentrations and were indistinguishable whether the bacterial culture was preincubated with the peptide or not.

The results obtained with ciprofloxacin and E. coli 045-849 indicated a trend that fell between the results obtained with the β-lactam and aminoglycoside antibiotics (Fig. 2D). Preincubation of bacteria with 8 μg/ml peptide A3-APO in the synergy lane did result in absorbance readings lower than those observed after ciprofloxacin or A3-APO administration alone, and the effect was not seen with coadministration of the two antibiotics. A lower peptide concentration (2 μg/ml) did not result in a dramatic recovery of ciprofloxacin activity, although some synergy was seen with or without preincubation, suggesting that the peptide derivative potentiated the activity of ciprofloxacin by eliminating a fraction of the induced DnaK population.

Taken together, addition of a pyrrhocoricin-based dimer recovered the activity of amoxicillin, a β-lactam antibiotic, against resistant strains. The synergy mode of action is likely inhibition of protein folding because no such effect could be observed with kanamycin, to which resistance involves changes in the target protein sequence. Peptide activity on TEM-1 inactivation appears to be a transient phenomenon. Upon A3-APO treatment of E. coli BF1023 or SEQ102, the cell pellet weights showed a time-dependent reduction, consistent with the kinetics of killing (7), but without a noticeable change in TEM-1 activity on nitrocefin after peptide removal and cell resuspension. This hypothesis is supported by reports of a reversible unfolding process (25) and nonpeptide inactivation of TEM-1 (27). As opposed to current β-lactamase inhibitors and combination therapies that work against a limited number of strains, inhibition of all protein folding in bacteria is a universal treatment option. Elimination of resistance to β-lactams by proline-rich peptide derivatives may give renewed life to these antibiotics, large stockpiles of which are currently available.

• Open in new tab
• Download powerpoint

FIG. 1.

β-Lactamase expression in the amoxicillin-resistant test bacterial strains. Isoelectric focusing of a standard containing a mixture of potential β-lactam resistance markers (lane Std) and Escherichia coli SEQ102 (lane 1) or BF1023 (lane 2) indicates intense TEM-1 production by the amoxicillin-resistant strains used in the current study.

• Open in new tab
• Download powerpoint

FIG. 2.

Killing of Escherichia coli by antibiotics, as represented by a broth microdilution liquid growth inhibition assay. Peptide A3-APO (A to D) or the small-molecule antibiotics were added to bacterial cultures alone in the 0.5- to 64-μg/ml concentration range. (A) E. coli BF1023-amoxicillin; (B) E. coli SEQ102-amoxicillin; (C) E. coli S5081-kanamycin; (D) E. coli 045-849-ciprofloxacin. Synergy assays were run on the same plates, where peptide A3-APO was added at two fixed concentrations either 1 h prior to small-molecule antibiotic addition or together with various concentrations of the conventional antibiotics.

• American Society for Microbiology