Novel Bacterial Topoisomerase Inhibitors with Potent Broad-Spectrum Activity against Drug-Resistant Bacteria

ABSTRACT

The novel bacterial topoisomerase inhibitor class is an investigational type of antibacterial inhibitor of DNA gyrase and topoisomerase IV that does not have cross-resistance with the quinolones. Here, we report the evaluation of the in vitro properties of a new series of this type of small molecule. Exemplar compounds selectively and potently inhibited the catalytic activities of Escherichia coli DNA gyrase and topoisomerase IV but did not block the DNA breakage-reunion step. Compounds showed broad-spectrum inhibitory activity against a wide range of Gram-positive and Gram-negative pathogens, including biodefence microorganisms and Mycobacterium tuberculosis. No cross-resistance with fluoroquinolone-resistant Staphylococcus aureus and E. coli isolates was observed. Measured MIC₉₀ values were 4 and 8 μg/ml against a panel of contemporary multidrug-resistant isolates of Acinetobacter baumannii and E. coli, respectively. In addition, representative compounds exhibited greater antibacterial potency than the quinolones against obligate anaerobic species. Spontaneous mutation rates were low, with frequencies of resistance typically <10⁻⁸ against E. coli and A. baumannii at concentrations equivalent to 4-fold the MIC. Compound-resistant E. coli mutants that were isolated following serial passage were characterized by whole-genome sequencing and carried a single Arg38Leu amino acid substitution in the GyrA subunit of DNA gyrase. Preliminary in vitro safety data indicate that the series shows a promising therapeutic index and potential for low human ether-a-go-go-related gene (hERG) inhibition (50% inhibitory concentration [IC₅₀], >100 μM). In summary, the compounds' distinct mechanism of action relative to the fluoroquinolones, whole-cell potency, low potential for resistance development, and favorable in vitro safety profile warrant their continued investigation as potential broad-spectrum antibacterial agents.

INTRODUCTION

Bacterial infections are becoming increasingly untreatable because of the rapid emergence of multidrug resistance as well as the limited number of novel antibacterial agents in clinical development (1–3). The U.S. Centers for Disease Control and Prevention (CDC) recently identified 15 antibiotic-resistant microorganisms as posing a threat to human health and classified as “urgent” or “serious” (4). Prominent among this set are antibiotic-resistant strains of the ESKAPE (Enterococcus faecium, Staphylococcus aureus, Klebsiella pneumoniae, Acinetobacter baumannii, Pseudomonas aeruginosa, and Enterobacter species) pathogens (1), such as carbapenem-resistant Enterobacteriaceae (CRE), multidrug-resistant (MDR) Acinetobacter, MDR Pseudomonas aeruginosa, methicillin-resistant Staphylococcus aureus (MRSA), and vancomycin-resistant Enterococcus (VRE). Also in the list are the Gram-positive anaerobe Clostridium difficile, drug-resistant Neisseria gonorrhoeae, and drug-resistant tuberculosis.

The urgent need to discover and develop new antibacterial agents to counter the threat of drug-resistant infections is widely recognized. Research efforts over the past few years have focused on the development of novel classes of antibacterials with a dual-targeting mechanism of action that is distinct from currently used antibiotics, with the twin objectives of avoiding cross-resistance and reducing the emergence of de novo resistance. The essential bacterial type II topoisomerase enzymes, DNA gyrase, and topoisomerase IV are well-validated drug targets for antibiotic pharmacology as evidenced by the fluoroquinolone and aminocoumarin classes of antibiotics (5–8). These enzymes are responsible for introducing negative supercoils into DNA and for the decatenation of DNA. The high degree of sequence similarity between DNA gyrase and topoisomerase IV offers the prospect of multitargeting with a single pharmacophore (9, 10). Despite the now widespread resistance to the quinolones, the type II topoisomerases continue to provide opportunities for antibacterial discovery based on exploiting novel binding interactions between new chemical ligands and the target enzymes in order to bypass mutations associated with quinolone resistance. Selected examples of this strategy are the 2-aminoquinazolinedione (11), the isothiazoloquinolone (12), the spiropyrimidinetrione (13), and the novel tricyclic topoisomerase inhibitor (NTTI) (14) classes.

One emerging class of non-quinolone inhibitors of DNA gyrase and topoisomerase IV is the novel bacterial topoisomerase inhibitor (NBTI) type. NBTI molecules bind to a site that is distinct from, but adjacent to, the catalytic center of DNA gyrase/topoisomerase IV, which is occupied by the quinolones (15). Consequently, NBTI compounds retain potency against fluoroquinolone-resistant (FQR) isolates. Structurally, NBTI molecules comprise a northern head group that interacts with the DNA, a central linker portion, and a southern group that binds to the enzymes. A number of advanced NBTI molecules have been described in the literature, including NXL101 (16), AZD9742 (17), NBTI 5463 (18), and gepotidacin (19), which recently successfully completed phase II human clinical evaluation for the treatment of uncomplicated urogenital gonorrhea caused by Neisseria gonorrhoeae (ClinicalTrials registration number NCT02294682). The NBTI pharmacophore, however, has been associated with cardiovascular and other safety liabilities (17, 20–23). Therefore, a key aim in the development of NBTIs is achieving broad antibacterial potency, including against challenging Gram-negative pathogens, while maintaining satisfactory safety margins.

Toward this goal, Redx Pharma recently disclosed a new series of NBTI type compounds characterized by a novel tricyclic northern head group as described in international patent WO 2016/024096 (24). The chemical structures of six selected compounds from this series are displayed in Fig. 1. The purpose of this study was to undertake a detailed in vitro biological evaluation of exemplar compounds from the series. Specifically, their ability to inhibit DNA gyrase and topoisomerase IV activities; their whole-cell potency against panels of wild-type (WT) and FQR bacteria, including clinically important anaerobes and biodefence organisms; and their in vitro safety profiles were assessed and are reported.

FIG 1.

Chemical structures of the compounds described in this study.

RESULTS

Inhibition of target activity in vitro.

The five compounds tested potently inhibited both Escherichia coli DNA gyrase and topoisomerase IV enzymes, consistent with a dual-targeting mechanism of action (Table 1). Ciprofloxacin was selected as a representative quinolone and tested in parallel for comparison. REDX05777, REDX06181, REDX06213, REDX07623, and REDX07638 produced a range of 50% inhibitory concentrations [IC₅₀s] comparable with ciprofloxacin in the supercoiling assay, while all five compounds showed approximately 10-fold lower IC₅₀s than ciprofloxacin in the decatenation assay. Stabilization of the DNA gyrase cleavage complex was observed in the presence of ciprofloxacin (35% at 100 μM), while all Redx compounds showed little or no stabilization of this complex at the same concentration. Taken together, these results indicate that the Redx compounds potentially have more balanced dual-targeting activity than ciprofloxacin and a distinct mechanism of action. Similar to ciprofloxacin, Redx compounds showed a high degree of selectivity for the bacterial enzymes over the homologous mammalian enzyme, human topoisomerase II, with approximately 2 orders of magnitude difference in measured IC₅₀s (Table 1).

TABLE 1.

Inhibition of the DNA supercoiling and cleaved complex formation activities of E. coli DNA gyrase (WT and Arg38Leu mutant), the decatenation activity of E. coli topoisomerase IV, and the decatenation activity of human topoisomerase II by ciprofloxacin and exemplar Redx NBTI compounds

Antibacterial compound	E. coli DNA gyrase (WT) IC50 (μM)	E. coli DNA gyrase (Arg38Leu mutant) IC50 (μM)	E. coli DNA gyrase cleavage complex (% cleaved at 100 μM)	E. coli topoisomerase IV IC50 (μM)	Human topoisomerase II IC50 (μM)
Ciprofloxacin	0.77	1.32	35.00	10.20	500
REDX05777	0.29	NDa	0	0.25	>100
REDX06181	1.47	21.80	2.60	1.17	84
REDX06213	1.66	1.83	3.50	0.14	100
REDX07623	0.21	3.57	0	0.10	>100
REDX07638	0.23	2.31	0	0.10	100

^aND, not determined.

Bacterial susceptibility profile.

Bacterial susceptibility profiling of the NBTI compounds shows that the series has broad-spectrum activity against the Gram-negative and Gram-positive pathogens tested, including those of the ESKAPE group of microorganisms (Table 2). Compounds from the series were generally more potent against Gram-positive species. Compounds were active against fastidious Gram-negative species, such as Haemophilus influenzae and Neisseria gonorrhoeae, as well as nonfastidious Gram-negative species. Of the nonfastidious Gram-negative species, Redx compounds demonstrated more potent activity against Acinetobacter baumannii and E. coli in comparison to that against Enterobacter cloacae, Klebsiella pneumoniae, and Pseudomonas aeruginosa. Representative compound REDX05777 also potently inhibited the whole-cell proliferation of Mycobacterium tuberculosis (Table 2). The exemplar compound REDX07638 was tested against a set of five biothreat microorganisms that included the etiological agents of anthrax, glanders, melioidosis, tularemia, and the plague (Table 3). REDX07638 inhibited all five Gram-positive and Gram-negative species, inhibiting Bacillus anthracis, Francisella tularensis, and Yersinia pestis most potently.

TABLE 2.

Bacterial susceptibility profile of NBTI compounds against reference bacterial strains and FQR mutants

Species and strain	MIC (μg/ml)
	Ciprofloxacin	REDX 05777	REDX 06181	REDX 06213	REDX 06276	REDX 07623	REDX 07638
Acinetobacter baumannii NCTC 13420	64	2	16	0.12	0.25	0.25	0.25
Clostridium difficile ATCC 700557	16	NDa	ND	2	ND	2	ND
Enterobacter cloacae NCTC 13406	0.015	8	8	2	1	2	2
Enterococcus faecalis ATCC 29212	1	2	2	1	0.5	1	1
Enterococcus faecium ATCC 19434	8	4	8	2	2	2	2
Escherichia coli ATCC 25922	0.03	0.5	1	0.12	0.25	0.5	0.5
E. coli MG1655 WT	0.008	0.5	1	0.25	0.5	0.5	0.5
E. coli MG1655 S83L	0.12	0.5	0.5	0.25	0.25	0.25	0.25
E. coli MG1655 D87G	0.06	1	0.5	0.5	1	1	1
E. coli ECCPX1-SP25	32	4	1	1	0.5	1	2
Haemophilus influenzae ATCC 49247	0.008	4	4	2	2	4	2
Klebsiella pneumoniae ATCC 700603	0.25	16	32	8	8	8	8
Mycobacterium tuberculosis H37Rv	2.2	1.3	ND	ND	ND	ND	ND
Neisseria gonorrhoeae ATCC 49226	0.004	4	16	2	1	2	2
Pseudomonas aeruginosa ATCC 27853	1	8	8	4	4	8	4
Staphylococcus aureus ATCC 29213	0.25	1	4	0.12	0.25	0.5	0.12
Streptococcus pneumoniae ATCC 49619	0.5	2	4	0.5	0.25	0.5	0.25

^aND, not determined.

TABLE 3.

Susceptibility profile of a panel of biodefence pathogens to REDX07638 and the comparator antibiotic doxycycline

Species and strain	MIC (μg/ml) of:
	Doxycycline	REDX07638
Bacillus anthracis Ames	0.03	0.5
Burkholderia mallei China 7	0.12	8
Burkholderia pseudomallei DD503	2	32
Burkholderia pseudomallei K96243	2	32
Burkholderia pseudomallei 1026b	1	16
Francisella tularensis SCHU S4	0.5	4
Yersinia pestis CO92	2	1

Compounds retained potency against the E. coli strains, MG1655 S83L and MG1655 D87G, carrying the Ser83Leu and Asp87Gly mutations in the GyrA subunit of DNA gyrase that are associated with quinolone resistance. In all cases, the MIC for the mutant was within 1 doubling dilution on either side of the MIC for the isogenic parent strain E. coli MG1655. In contrast, the MIC of ciprofloxacin increased 8- to 16-fold against the MG1655 S83L and MG1655 D87G strains compared to that of the isogenic parent (Table 2). E. coli ECCPX1-SP25 is a ciprofloxacin-resistant mutant derived from strain ATCC 25922. The MIC of ciprofloxacin against this strain is elevated 1,024-fold relative to that of the parent strain. In comparison, the activity of Redx compounds against E. coli ECCPX1-SP25 was within 2- to 8-fold of the MIC of strain ATCC 25922 (Table 2). Taken together, these results indicate a lack of cross-resistance of the NBTI series with the quinolone class of antibiotics.

A selection of Redx compounds was tested against a panel of recent MDR and FQR Gram-negative clinical isolates (Table 4). All three compounds tested showed antibacterial activity against E. coli with MIC₉₀ values of 4 or 8 μg/ml. Similar activity was observed with REDX07623, REDX06213, and REDX06276 against A. baumannii with MIC₉₀ values of 4 or 8 μg/ml. The MIC₉₀ values observed for these NBTI compounds against E. coli and A. baumannii were lower than those obtained for the fluoroquinolone antibiotic levofloxacin (16 μg/ml). REDX06276 was the most active compound from this series against the K. pneumoniae panel, with a MIC₉₀ of 16 μg/ml, which is comparable to levofloxacin. Activity against P. aeruginosa and E. cloacae was observed at 32 to 64 μg/ml for all of the compounds tested, which was similar to the MIC₉₀ values obtained for levofloxacin.

TABLE 4.

MIC₉₀ of NBTI compounds and levofloxacin for a panel of recently isolated levofloxacin-resistant and MDR clinical isolates

Species (no. of isolates)	MIC90 (μg/ml [range]) of:
	Levofloxacin	REDX06213	REDX06276	REDX07623
K. pneumoniae (42)	16 (0.03 to >64)	32 (2 to >64)	16 (2 to >64)	32 (4 to >64)
A. baumannii (43)	16 (0.06 to >64)	8 (1 to 16)	8 (0.5 to 16)	4 (0.5 to 32)
P. aeruginosa (42)	64 (0.03 to >64)	32 (2 to >64)	32 (2 to 64)	32 (4 to >64)
E. cloacae (41)	32 (0.03 to >64)	32 (4 to >64)	32 (4 to >64)	32 (4 to >64)
E. coli (43)	16 (0.03 to >64)	4 (1 to >64)	4 (0.5 to >64)	8 (1 to >64)

Finally, compounds were tested for activity against species of clinically significant obligate anaerobic Gram-positive and Gram-negative bacteria. Antibiotics of the quinolone class have generally shown poor to moderate in vitro antibacterial potency against anaerobic bacteria relative to that of other classes of antibiotics and compared with their potency against aerobic bacteria (25, 26). Metronidazole and vancomycin were equally effective against a panel of recently isolated anaerobes, including the Gram-positive strains of Peptoniphilus harei, Peptostreptococcus anaerobius, and Clostridium perfringens with MIC₉₀ values lower or equal to 2 μg/ml (Table 5). Metronidazole, however, was not active against Propionibacterium acnes while the Redx compounds maintained activity with MIC₉₀ values of 0.5 and 4 μg/ml. Similar to ciprofloxacin, the tested compounds showed reduced activity against the Gram-positive strains of Finegoldia magna and Parvimonas micra with MIC₉₀ values of 16 to 64 μg/ml. Redx compounds showed activity at 2 to 16 μg/ml against the four Gram-negative bacterial species tested. Although metronidazole displayed lower MIC₉₀ values (0.25 to 2 μg/ml), the NBTI compounds showed improved activity compared to that of ciprofloxacin and vancomycin.

TABLE 5.

Bacterial susceptibility profile of NBTI compounds and comparator antibiotics against a panel of recent clinical isolates of 10 anaerobic bacterial species

Species (no. of isolates)	MIC90 (μg/ml [range]) of:
	Metronidazole	Vancomycin	Ciprofloxacin	REDX06213	REDX07623
Clostridium perfringens (11)	1 (0.25 to 1)	0.5 (0.5)	0.5 (0.25 to 0.5)	2 (0.25 to 2)	2 (0.5 to 4)
Finegoldia magna (12)	0.5 (0.5 to 1)	0.5 (0.25 to 0.5)	32 (0.25 to 32)	16 (1 to >64)	32 (0.5 to >64)
Parvimonas micra (12)	1 (0.25 to 1)	1 (≤0.12 to 4)	16 (0.5 to 16)	64 (2 to 64)	64 (0.5 to 64)
Peptostreptococcus anaerobius (10)	0.5 (0.25 to 0.5)	0.5 (0.25 to >64)	16 (0.5 to 16)	0.25 (≤0.12 to 8)	1 (≤0.12 to 8)
Propionibacterium acnes (12)	>32 (>32)	0.5 (0.25 to 16)	1 (0.25 to 2)	0.5 (≤0.12 to 0.5)	4 (≤0.12 to 4)
Peptoniphilus harei (12)	1 (0.25 to 1)	0.12 (≤0.12)	4 (1 to 16)	≤0.12 (≤0.12 to 0.25)	0.5 (≤0.12 to 0.25)
Bacteroides fragilis (12)	0.5 (0.25 to 0.5)	32 (16 to 32)	16 (2 to 32)	4 0.(25 to 64)	4 (0.5 to 64)
Bacteroides thetaiotaomicron (11)	1 (0.5 to 1)	64 (0.5 to 64)	64 (0.5 to 64)	16 (0.25 to 16)	8 (1 to 8)
Prevotella bivia (10)	2 (1 to 2)	>64 (32 to >64)	32 (8 to 64)	8 (2 to 8)	8 (2 to 8)
Prevotella melaninogenica (11)	0.25 (0.06 to 0.5)	>64 (8 to >64)	4 (0.5 to 8)	2 (≤0.12 to 4)	4 (≤0.12 to 4)

Time-kill.

The rate of bactericidal activity was determined for REDX06213 and REDX07623 against A. baumannii NCTC 13420 at 4× and 8× MIC. REDX06213 demonstrated bactericidal activity, showing a 3-log drop in CFU per milliliter at 2.6 and 2.8 h at 4× and 8× MIC, respectively (Fig. 2). A 3-log reduction in CFU was not achieved within 24 h with REDX07623 at 4× MIC; however, at 8× MIC, a 3-log drop in CFU per milliliter was observed at 0.97 h (Fig. 2). Regrowth was observed with REDX07623 at 4× and 8× MIC and with REDX06213 at 4× MIC but not at 8× MIC. Regrowth at 24 h is not uncommon and has previously been reported for bactericidal antimicrobials, such as ciprofloxacin against E. coli (27). However, this effect was not reported in a separate study with A. baumannii (28).

FIG 2.

Bactericidal activity of REDX06213 and REDX7623 at 4× and 8× MIC against A. baumannii NCTC 13420. Closed diamonds, drug-free control; closed squares, Redx compound at 4× MIC; closed triangles, Redx compound at 8× MIC.

Selection of resistant mutants.

In order to assess the propensity for the development of de novo resistance to this class of NBTI compounds, the spontaneous frequency of resistance (FoR) to REDX06213, REDX06276, REDX07623, and REDX07638 was determined with E. coli strain ATCC 25922. No mutants could be isolated at concentrations equivalent to 4× MIC, yielding a frequency of resistance ranging from <2.5 × 10⁻⁹ to <3.3 × 10⁻⁹. By comparison, the frequency of resistance to ciprofloxacin at 4× MIC was 7.8 × 10⁻⁸ for E. coli ATCC 25922. To confirm that the observed mutation frequencies were not species specific, frequency of resistance values were also determined for REDX06213, REDX07623, and REDX07638 in A. baumannii strain NCTC 13420. Again, no mutants were isolated, yielding frequency of resistance values between <6.7 × 10⁻⁸ and <7.4 × 10⁻⁸.

Next, E. coli ATCC 25922 was used in serial passage experiments with REDX06276 as a representative compound from this NBTI series. Ciprofloxacin and delafloxacin were used as comparator antibiotics. The MIC of ciprofloxacin increased up to 64 μg/ml after 25 passages with resistance observed at passage 23 (MIC, ≥4 μg/ml). The MIC of delafloxacin, however, remained within 2-fold of the original MIC (0.5 μg/ml) up to passage 24, after which it increased steadily to reach 16 μg/ml (32-fold increase) at passage 45. The MIC of REDX06276 followed a trend comparable to delafloxacin with an increase up to 32-fold (MIC, 64 μg/ml) at passage 45, at which stage the experiment was ended (Fig. 3). Whole-genome sequencing of the ciprofloxacin-resistant mutant from passage 25 (ECCPX1-SP25) revealed Ser83Leu and Asp87Gly mutations in the GyrA subunit and a Glu84Lys mutation in the ParC subunit. The delafloxacin-resistant mutant at passage 45 had target gene mutations corresponding to Ala119Glu and Ala179Val amino acid substitutions in the GyrA subunit. The REDX06276-resistant mutant from passage 45 carried a single Arg38Leu substitution in the GyrA subunit. In addition to target mutations, several off-target mutations were identified in all mutants (see Table S1 in the supplemental material). Introduction of the single Arg38Leu mutation into E. coli DNA gyrase led to a modest increase in IC₅₀ of 10- to 17-fold for all compounds, with the exception of REDX06213 (no significant change in activity) in the enzyme assay (Table 1). However, whole-cell MIC testing revealed a loss in potency of 64- to 256-fold (see Table S2 in the supplemental material).

FIG 3.

Isolation of drug-resistant mutants of E. coli ATCC 25922 by serial passage. Closed circles, REDX06276; closed triangles, ciprofloxacin; closed squares, delafloxacin.

In vitro safety profile.

Mammalian cytotoxicity testing with the HepG2 cell line revealed IC₅₀s that were higher than the corresponding MIC values observed with the ESKAPE pathogens by more than 2 orders of magnitude (Table 6). In vitro testing showed the series to have a range of human ether-a-go-go-related gene (hERG) inhibitory activity, with REDX07623 having an IC₅₀ of 8.2 μM, while REDX6181 demonstrated reduced hERG inhibition with an IC₅₀ of >100 μM. A trend between logD and hERG activity was found with this series. Compounds with a lower logD (distribution coefficient) appeared to have reduced hERG inhibition (Table 6).

TABLE 6.

In vitro safety profile of NBTI compounds

Compound	HepG2 IC50 (μg/ml)	hERG IC50 (μM)	logD
REDX05777	>64	>33	0.4
REDX06181	>64	>100	−0.64
REDX06213	29.6	>33	0.94
REDX06276	>64	>33	0.71
REDX07623	>64	8.2	1.26
REDX07638	38	8.9	1.39

DISCUSSION

In recent years, the growing threat of drug-resistant bacterial infections, combined with the lack of new antibiotics with a novel mechanism of action, has caused global concern. Resistance of Gram-negative species to first-line and last-resort antibiotics has been reported worldwide and can lead to untreatable infections and increased mortality (29). To address this unmet medical need, this study describes the in vitro assessment of a NBTI series with dual-targeting activity against bacterial DNA gyrase and topoisomerase IV and with a mechanism of action different from clinically used fluoroquinolones. Redx compounds demonstrated potent, balanced inhibitory activity versus the two topoisomerase enzymes, with IC₅₀s ranging from 0.21 to 1.66 μM against E. coli DNA gyrase and from 0.10 to 1.17 μM against E. coli topoisomerase IV (Table 1). Inhibitory activity was more balanced than ciprofloxacin, which had IC₅₀s of 0.77 and 10.20 μM against E. coli gyrase and topoisomerase IV, respectively. This is in agreement with data recorded in the literature, which show ciprofloxacin to have a preference for DNA gyrase in E. coli and topoisomerase IV in S. aureus (30). Importantly, selectivity over human topoisomerase II was found with bacterial enzymes showing an approximate 100-fold increase in sensitivity to Redx compounds. The formation of DNA-cleaved complexes was limited with Redx compounds in comparison to the level observed with ciprofloxacin (Table 1). This indicates that the NBTI series described here has a mechanism of action different from ciprofloxacin, which stabilizes double-stranded broken DNA strands, blocking re-ligation, the consequences of which are poisonous to the bacterial cell. Instead, the NBTIs described here interact with the topoisomerase and DNA prior to double-strand breakage, which has been reported for other NBTI series (18).

Broad-spectrum antibacterial activity was found with this series against a panel of ESKAPE pathogens, the fastidious Gram-negative organisms H. influenzae and N. gonorrhoeae, as well as M. tuberculosis, and important Gram-positive and Gram-negative biothreat pathogens (Table 2 and 3). Potency was maintained against FQR E. coli isolates with a single amino acid substitution in the GyrA subunit (Ser83Leu or Asp87Gly) with MICs within 2-fold of the MIC for the isogenic parent strain. Additionally, compounds retained potency against the serial passage FQR mutant E. coli ECCPX1-SP25, with MICs increasing 8-fold or less compared to that of ciprofloxacin, which showed a 1,024-fold increase in MIC. The retained potency of this series against FQR mutants supports that the mechanism of action is different from the fluoroquinolones indicated by the cleavage complex enzyme assay. Overall, a reduction in antibacterial activity was found when Redx compounds were tested against a larger panel of clinically relevant strains (25% MDR). Antibacterial activity was retained against A. baumannii and E. coli, with values of 4 to 8 μg/ml; however, a loss of potency was found against the other Gram-negative species, K. pneumoniae, E. cloacae, and P. aeruginosa, with MIC₉₀ values of 16 to 32 μg/ml. The activity of Redx compounds was equal or superior to that of levofloxacin and other fluoroquinolones reported in the literature (31); however, the reduced activity against a wider panel of strains shows modifications to improve antibacterial potency will be essential during continued development. In addition to good activity against ESKAPE pathogens, compounds demonstrated antibacterial activity against a panel of anaerobic pathogens, including Clostridium and Bacteroides species (Tables 2 and 5). In Europe, C. difficile is estimated to cause 250,000 infections and 14,000 deaths per annum, showing resistance to a large number of antibiotics, including the fluoroquinolones (32). Bacteroides species are part of the mammalian gut microbiota and can be opportunistic pathogens as well as a reservoir for resistance. They are also frequently resistant to a wide range of antibiotics, necessitating the development of novel compounds that are effective against these species.

Rapid bactericidal activity of the series was demonstrated with representative compounds, REDX06213 and REDX07623, against A. baumannii NCTC 13420, with both compounds causing a 3-log drop in CFU per milliliter at 2.6 h (4× MIC) and 0.97 h (8× MIC), respectively (Fig. 2). The rate of bactericidal activity was similar or superior to that for ciprofloxacin and other NBTIs reported in the literature (27, 33, 34). Interestingly, REDX06213 caused a similar rate of bactericidal activity at 4× and 8× MIC, indicating time-dependent rather than concentration-dependent killing.

No mutants were raised against compounds tested at 4× MIC with A. baumannii NCTC 13420 and E. coli ATCC 25922. In contrast, a mutation rate of 4.76 × 10⁻⁸ was obtained with E. coli ATCC 25922 against ciprofloxacin at equivalent multiples of its MIC. These results indicate a low potential for resistance development to this series and support the balanced dual-targeting activity revealed by the supercoiling and decatenation assay data. Development of resistance to other NBTI series at 4× MIC has been reported in the literature. Resistance rates of 5 × 10⁻⁸ were found with NBTI 5463 against P. aeruginosa PAO1, although sequencing showed no target gene mutations (18). Second- and third-step mutants had Asp82Glu and Asp82Glu plus Asp87Tyr point mutations in GyrA, respectively (35). These mutants showed no cross-resistance to the fluoroquinolones, which is consistent with a differential binding mechanism between the NBTI and fluoroquinolone classes of topoisomerase inhibitors. Although no mutants were raised during the spontaneous frequency of resistance experiments against the compounds described here, whole-genome sequencing of the E. coli REDX06276 serial passage mutant revealed a single Arg38Leu substitution in the GyrA subunit. This mutation has been reported previously and conferred resistance to 5,6-bridged quinolones but not to other quinolones (36). Similarly, no cross-resistance to ciprofloxacin was found with the E. coli REDX06276 serial passage mutant (see Table S2 in the supplemental material). To understand the contribution of the GyrA Arg38Leu substitution to the increased resistance of E. coli REDX06276, 4 Redx compounds were tested against the mutant gyrase in the supercoiling assay (Table 1). Introduction of the Arg38Leu mutation into the wild-type enzyme led to a modest increase in IC₅₀ of 10- to 17-fold, except for REDX06213, which showed no significant change in activity. Whole-cell MIC testing revealed a loss in potency of 64- to 256-fold (Table S2), suggesting that additional off-target mutations may be contributing to the resistance observed in this strain. Indeed, a mutation in the efflux repressor gene acrR may have also contributed to increased resistance through attenuated suppression of acrB expression.

Compounds in this series show a promising safety profile with HepG2 cytotoxicity IC₅₀s of ≥32 μg/ml. Low hERG inhibition with the series has also been demonstrated, with IC₅₀s of >100 μM. During the optimization of this series, efforts have been made to reduce hERG channel inhibition while retaining antibacterial potency. The addition of a fluorine atom in the southern group of REDX07623 appears to increase hERG affinity in comparison to that in its matched pair REX06276 with IC₅₀s of 8.2 and >33 μM, respectively. Introduction of more polar groups to reduce the logD of NBTI compounds has been shown to be associated with reduced hERG inhibition (22). REDX06181 displayed the lowest logD of the compounds tested and showed the most attenuated hERG inhibition with an IC₅₀ of >100 μM. However, its antibacterial potency was reduced compared to that of other compounds with a higher logD, such as REDX07623. A negative correlation between whole-cell antibacterial potency and hERG inhibition has been reported for other NBTI type compounds (22).

In summary, the NBTI series described here shows potent, balanced, dual-targeting inhibition of DNA gyrase and topoisomerase IV, with selectivity over human topoisomerase II. Data from DNA-cleaved complex experiments indicate that the series has a mechanism of action different from the fluoroquinolones. The low mutation rate of Gram-negative strains to the compounds combined with the balanced inhibitory enzyme activity suggests that resistance may be slow to develop during therapeutic use. Antibacterial activity was demonstrated against a wide panel of susceptible and drug-resistant bacterial species, including the ESKAPE set of organisms, medically important anaerobic species, and other pathogens, including larger sets of MDR isolates. Rapid bactericidal activity was also demonstrated. These properties, in combination with the promising in vitro safety profile, warrant the further development of this NBTI series.

MATERIALS AND METHODS

Reagents and media.

Proprietary compounds were prepared at Redx Pharma as described in the international patent WO 2016/024096. Reference antibiotics were purchased from Sigma-Aldrich (Dorset, United Kingdom). Bacteriological media were purchased from Oxoid Ltd. (Basingstoke, United Kingdom).

Bacterial strains.

The bacteria used in this study were obtained from the American Type Culture Collection (ATCC; Middlesex, United Kingdom), the Network on Antibacterial Resistance in Staphylococcus aureus (Manassas, VA), or the Coli Genetic Stock Center (New Haven, CT). Escherichia coli strains MG1655 WT, MG1655 S83L, and MG1655 D87G were provided by T. Maxwell (John Innes Centre, Norwich, United Kingdom). E. coli ECCPX1-SP25 was selected and characterized at Redx Pharma by the serial passage of E. coli ATCC 25922 in the presence of ciprofloxacin as described in international patent WO 2016/024098 (37).

DNA supercoiling, decatenation, and cleavage complex.

DNA supercoiling, decatenation, and cleavage complex assays were all performed by Inspiralis Ltd. (Norwich, United Kingdom) using a gel-based assay format. Briefly, 1 unit of E. coli DNA gyrase (WT or Arg38Leu mutant) was incubated with 0.5 μg of relaxed pBR322 and 1 unit of topoisomerase IV (WT), or human topoisomerase II was incubated with 200 ng kinetoplast DNA (kDNA), all in a reaction mixture volume of 30 μl at 37°C for 30 min in the presence of a series of concentrations of the test compound. Supercoiling reactions were conducted under the following conditions: 35 mM Tris-HCl (pH 7.5), 24 mM KCl, 4 mM MgCl₂, 2 mM dithiothreitol (DTT), 1.8 mM spermidine, 1 mM ATP, 6.5% (wt/vol) glycerol, and 0.1 mg/ml bovine serum albumin (BSA). E. coli topoisomerase IV decatenation reactions were conducted under the following conditions: 50 mM HEPES-KOH (pH 7.6), 100 mM potassium glutamate, 10 mM magnesium acetate, 10 mM dithiothreitol, 1 mM ATP, and 50 μg/ml BSA. Inhibition of human topoisomerase II decatenation activity was assessed as described previously (14). Reactions were stopped using 30 μl chloroform/isoamyl alcohol (26:1) and 20 μl stop dye (40% sucrose, 100 mM Tris-HCl [pH 7.5], 1 mM EDTA, 0.5 μg/ml bromophenol blue). Topoisomers were visualized by ethidium bromide staining, resolved, and quantified by gel electrophoresis, and the band intensities were analyzed by gel documentation equipment (Syngene, Cambridge, United Kingdom) and quantified using Syngene GeneTools software. Raw data were converted to a percentage of the inhibitor-free control and were analyzed using SigmaPlot version 12.5. Nonlinear regression was used to calculate the 50% inhibitory concentrations (IC₅₀s). The human topoisomerase II inhibitor, etoposide, was used as a positive control for inhibition for this assay. For cleavage complex assays, compounds were tested at 100 μM in a final dimethyl sulfoxide (DMSO) concentration of 1% (vol/vol). E. coli DNA gyrase (1 unit) was incubated with 0.5 μg of supercoiled pBR322 DNA at 37°C for 30 min. Reactions were performed in a volume of 30 μl using the following conditions: 35 mM Tris-HCl (pH 7.5), 24 mM KCl, 4 mM MgCl₂, 2 mM DTT, 1.8 mM spermidine, 6.5% (wt/vol) glycerol, and 0.1 mg/ml BSA. Following this, reaction mixtures were incubated for 30 min with 0.2% SDS and 0.5 μg/μl proteinase K. Reactions were stopped in the same manner as for the supercoiling and decatenation assays. Topoisomers and cleavage products were visualized by gel electrophoresis. Cleavage products were expressed as a percentage of the fully supercoiled inhibitor-free control as described for the supercoiling and decatenation assays.

Antibacterial susceptibility testing.

MICs were determined by the broth microdilution procedure according to the guidelines of the Clinical and Laboratory Standards Institute document M07-A10 (38). The broth microdilution method involved a 2-fold serial dilution of compounds in 96-well microtiter plates, giving a typical final concentration range of 0.25 to 128 μg/ml and a maximum final concentration of 1% DMSO. Strains were grown in cation-adjusted Mueller-Hinton broth (CA-MHB) or agar (CA-MHA) with or without 5% lysed horse blood at 37°C in an ambient atmosphere, in Haemophilus testing medium broth at 37°C in an ambient atmosphere, or in gonococcal broth or agar supplemented with Vitox at 37°C in an atmosphere containing 5% CO₂. The MIC was determined as the lowest concentration of compound that inhibits visible growth following a 16- to 24-h incubation period. For Mycobacterium tuberculosis, a fluorescent reporter strain of H37Rv was used and the MIC was determined by measuring the optical density at 590 nm (OD₅₉₀) or fluorescence (excitation [Ex], 560 nm/emission [Em], 590 nm) after 5 days of growth in 7H9 broth with 10% (vol/vol) oleic acid-albumin-dextrose-catalase (OADC) supplement and 0.05% (wt/vol) Tween 80 in the presence of test compound with a final DMSO concentration of 2%. MIC₉₀ determination was performed at IHMA Europe Sàrl (Epalinges, Switzerland) with a selection of clinical isolates collected between 2012 and 2014. Bacteria were obtained from a variety of infection types and geographical locations, including at least 25% highly drug-resistant isolates (resistant to at least seven of the following: amikacin, aztreonam, cefepime, ceftazidime, ceftriaxone, colistin, gentamicin, imipenem/meropenem, levofloxacin, piperacillin-tazobactam, and tetracycline) and with a selection of 10 different species of anaerobes, including 113 isolates collected in 2015 from diverse geographical origins. MICs were determined using frozen 96-well antibacterial panels prepared by broth microdilution in line with the guidelines of Clinical and Laboratory Standards Institute document M11-A8 (39), giving a final compound concentration range of 0.004 to 64 μg/ml. The inoculum size was 5 × 10⁵ CFU/ml and 5 × 10⁷ CFU/ml for the aerobic and anaerobic strains, respectively. The testing plates for anaerobes were incubated for 48 h at 35°C with 5% CO₂ in an anaerobic cabinet (Whitley A35 anaerobic workstation; Don Whitley Scientific). MICs were read visually, and values were reported as MIC₉₀ for inhibition of 90% of the isolates. MIC testing of Bacillus anthracis, Burkholderia mallei, Burkholderia pseudomallei, Francisella tularensis, and Yersinia pestis was undertaken by Southern Research (Birmingham, AL). The assay medium was CA-MHB (supplemented with 2% IsoVitaleX in the case of F. tularensis). Cultures were incubated in the presence of compound for up to 24 h, up to 38 h (F. tularensis), or 24 to 48 h (Y. pestis).

Time-kill.

The rate of the bactericidal activity of compounds was determined against A. baumannii NCTC 13420 at 4× and 8× MIC according to the guidelines of the Clinical and Laboratory Standards Institute document M26-A (40). A. baumannii was cultured overnight at 37°C, diluted in fresh CA-MHB, and grown to exponential phase (OD₆₀₀ = 0.3). Cultures were then adjusted to a 0.5 McFarland standard (1 × 10⁸ to 2 × 10⁸ CFU/ml) and diluted to approximately 5 × 10⁵ CFU/ml before addition of compound to give a final concentration of 4× or 8× MIC. Samples were taken at 0, 0.5, 1, 3, 6, and 24 h, serially diluted, and plated onto MHA; this was followed by overnight incubation at 37°C. The following day, colonies were enumerated to determine CFU per milliliter.

Frequency of resistance.

Overnight cultures of bacteria were grown from single colonies in CA-MHB. The following day, samples of the neat cultures were spread onto CA-MHB containing compound at the concentrations indicated. To determine the number of viable cells in the inoculum, samples of the overnight cultures were serially diluted in phosphate-buffered saline (PBS) and plated on compound-free CA-MHA. Plates were incubated for up to 48 h, and the colonies were enumerated. The spontaneous frequency of resistance (FoR) was calculated by dividing the number of resistant colonies (CFU/ml) by the total number of viable cells (CFU/ml).

Selection of resistant mutants by serial passage.

Resistant mutants were selected by serial passage carried out using the broth microdilution method. Following MIC determination, the culture representing 0.25× MIC was used to inoculate the subsequent passage until the desired level of resistance was achieved. At this point, clones were isolated and the MIC was confirmed as described above.

Whole-genome sequencing.

Genomic DNA (gDNA) was extracted from the resistant strains using the PurElute bacterial genomic kit (Edge Biosystems, Gaithersburg, MD). The gDNA was purified according to the manufacturer's recommendations. Purified gDNA was used to create whole-genome libraries using the NEBNext Ultra kit, and 150-bp paired-end read sequence data were produced using an Illumina HiSeq 3000. Read data were stored as FASTQ files, and then adaptor sequences were removed using cutadapt software (version 1.8). Data for the wild-type strain were used to construct a reference genome sequence using the CLC Bio genome assembler (version 8.0.1). Sequence data for each sample, including the progenitor strain, were aligned to the published E. coli ATCC 25922 genome using BWA (version 0.7.12); aligned data were sorted using Samtools6 (version 1.2). Variants were identified using VarScan (version 2.3.7) using the E. coli ATCC 25922 assembled genome as the reference sequence. The resulting data provided coverage of >100 reads across the genome. Single nucleotide polymorphisms (SNPs), insertions, and deletions were identified that were prevalent in ≥95% of the reads compared with those of the progenitor strain.

Cytotoxicity testing.

HepG2 cells (ATCC HB-8065) were seeded at a density of 20,000 cells per well and incubated for 24 h at 37°C in an atmosphere of 5% CO₂. Cells were then exposed to a doubling dilution series of the test compound. After 24 h of incubation, the viability of the cells was determined using CellTiter-Glo (Promega, WI, USA) according to the manufacturer's recommendations. Each experiment was carried out in duplicate, and the results were reported as the average concentration of test compound inhibiting 50% of cell viability (IC₅₀).

hERG inhibition.

Inhibition of the human ether-a-go-go-related gene (hERG) cardiac potassium (K⁺) ion channel was determined in a transfected Chinese Hamster Ovary K1 (CHO) cell line using IonWorks patch clamp electrophysiology (41).

LogD measurements.

Partitioning of compounds between 1-octanol and 0.1 M phosphate buffer (pH 7.4) was measured using the shake-flask method (42).