Discovery of a Potent Picolinamide Antibacterial Active against Clostridioides difficile

Abstract

A major challenge for chemotherapy of bacterial infections is perturbation of the intestinal microbiota. Clostridioides difficile is a Gram-positive bacterium of the gut that can thrive under this circumstance. Its production of dormant and antibiotic-impervious spores results in chronic disruption of normal gut flora and debilitating diarrhea and intestinal infection. C. difficile is responsible for 12,800 deaths per year in the United States. Here, we report the discovery of 2-(4-(3-(trifluoromethoxy)phenoxy)-picolinamido)benzo[d]oxazole-5-carboxylate as an antibacterial with potent and selective activity against C. difficile. Its MIC₅₀ and MIC₉₀ (the concentration required to inhibit the growth of 50% and 90% of all the tested strains, respectively) values, documented across 101 strains of C. difficile, are 0.12 and 0.25 μg/mL, respectively. The compound targets cell wall biosynthesis, as assessed by macromolecular biosynthesis assays and by scanning electron microscopy. Animals infected with a lethal dose of C. difficile and treated with compound 1 had a similar survival compared to treatment with vancomycin, which is the frontline antibiotic used for C. difficile infection.

Graphical Abstract

Clostridioides difficile is a Gram-positive, anaerobic, sporeforming bacterium. It is currently designated by the Centers for Disease Control and Prevention as an urgent health problem causing life-threatening diarrhea. An estimated 223,900 hospitalizations with C. difficile infection (CDI) occurred in the United States in 2017, with approximately 12,800 deaths.¹ One-third of the patients experience recurring infection leading to increased morbidity and mortality.² Furthermore, C. difficile produces spores that are resistant to antibiotics and can remain dormant for many months and years.^3,4 The disruption of the normal gut flora due to the use of broad-spectrum antibiotics facilitates the colonization and proliferation of C. difficile in the large intestine.⁵ Treatments for CDI include metronidazole (MTZ), which is no longer recommended as a first-line therapy for CDI, and vancomycin (VAN), which is currently the first-line option for CDI.⁶ However, MTZ is toxic, and it also kills beneficial gut bacteria. Fidaxomicin (FDX) is a narrow-spectrum bactericidal antibiotic effective against C. difficile; however, its use is limited due to its high cost and decreased efficacy against the highly virulent BI/NAP1/027 strain.⁷ A fecal microbiota transplant has been explored as CDI treatment. However, it is not standardized, and it has safety concerns such as the risk of transferring pathogenic organisms or autoimmune or metabolic disorders from donors to recipients.⁸ While a fecal microbiota transplant is effective, it has a failure of 10–20%.⁹ Several antimicrobial agents have been reported recently against C. difficile. The lipopeptide surotomycin and the oxazolidinone antibiotic cadazolid were evaluated in phase 3 clinical trials but were found to be noninferior to VAN.^10,11 Major challenges that have not been overcome for these agents are disruption of normal gut flora, the development of resistance, and a failure to reduce recurrence.¹²

We report the discovery of 2-(4-(3-(trifluoromethoxy)-phenoxy)picolinamido)benzo[d]oxazole-5-carboxylate (compound 1). This compound exhibits potent antibacterial activity against C. difficile without significant activity against other major gut bacteria. Compound 1 was identified in the course of research on the cinnamonitrile potentiators of β-lactam antibiotics against methicillin-resistant Staphylococcus aureus (MRSA).¹³ While the cinnamonitrile potentiators do not exhibit antibacterial activity, the insertion of an oxazole moiety into this structural template imparted antibacterial activity. Further structural elaboration of the picolinamide moiety significantly altered the structural template from the cinnamonitriles, leading to compound 1. Compound 1 is a potent and uniquely selective anti-C. difficile antibacterial.

Compound 1 was synthesized by the eight-step convergent synthesis shown in Scheme 1. 3-(Trifluoromethoxy)phenol (2) was allowed to react with 4-chloropicolinonitrile in the presence of potassium carbonate, followed by conversion of the nitrile in 3 to the carboxylic acid (4). Separately, we synthesized methyl 2-aminobenzo[d]oxazole-5-carboxylate (5, Scheme 1B). The reaction of cyanogen bromide (8) with imidazole produced di(1H-imidazol-1-yl)methanimine (9). The ring-closing double-nucleophilic attack of methyl 3-amino-4-hydroxybenzoate on 9 gave the desired intermediate 5. Coupling of 5 with 4 in the presence of propylphosphonic anhydride (T₃P) gave 6. Saponification of the ester moiety in 6 with sodium hydroxide and the subsequent work up gave the carboxylic acid 7. Compound 1 was obtained by the reaction of 7 with aqueous sodium bicarbonate.

Scheme 1.

Syntheses of Compounds (A) 1 and (B) 5

Compound 1 was evaluated against C. difficile ATCC 43255, a strain that produces toxins A (TcdA) and B (TcdB), both instigators for a bad outcome in CDI.¹⁴ The minimum inhibitory concentration (MIC) for 1 is 0.125 μg/mL. This value is 2-fold higher than FDX and 4-fold and 2-fold lower than VAN and MTZ, respectively (Table 1). The activity of compound 1 was evaluated against 101 C. difficile strains, including both clinical and laboratory strains, and VAN-, MTZ-, and FDX-resistant strains (Table S1). The MIC₅₀ and MIC₉₀ values (the concentration required to inhibit the growth of 50% and 90% of all the tested strains, respectively) are excellent at 0.12 and 0.25 μg/mL, respectively. As assessed by these MIC values, compound 1 has a similar potency as FDX but a better potency than the first-line antibiotics VAN and MTZ. The minimum bactericidal concentration (MBC) for 1 is 4 μg/mL or 64-fold higher than its MIC. Hence (by definition), 1 is bacteriostatic.¹⁵

Table 1.

MIC Values in μg/mL for Compound 1, VAN, MTZ, and FDX against C. difficile, MRSA, and Major Gut Bacteria^a

	1	VAN	MTZ	FDX
C. difficile ATCC43255b	0.125	0.5	0.25	0.06
C. difficile BAA-1870c	0.125	2	0.5	0.25
C. difficile 26675d	0.25	1	4	4
C. difficile 23828e	0.25	2	0.5	>8
C. difficile 23691f	0.125	4	0.5	4
MIC50 (101 strains)	0.12	0.5	0.5	0.06
MIC90 (101 strains)	0.25	2	1	0.25
S. aureus ATCC29213	64	1	>128	8
E.faecium NCTC7171	>128	0.5	>128	2
MRSA NRS70	16	2	>128	8
MRSA NRS119	16	2	>128	2
Bacteroides fragilisg	2	16	0.5	>32
Bifidobacterium longumh	16	0.25	0.5	<0.01
Corynebacterium spp.i	8	0.25	>32	<0.06
Fusobacterium nucleatumj	4	0.25	2	<0.06
Lactobacillus reuterik	4	>32	>32	>32
Lactobacillus gasseril	>16	1	>32	2
Veillonella sp. HM-49m	16	>32	2	8
Eubacterium sp.n	8	2	1	16
fecal microbiota preparationo	>128	>128	>128	>128

^aVAN: vancomycin. MTZ: metronidazole. FDX: fidaxomicin.

^bIsolated from abdominal wound, TcdA+ and TcdB+.

^cNAP1, BI 8, ribotype 27, toxinotype IIIb, TcdA+, TcdB+, and CDT+ (binary toxin).

^dClinical isolate resistant to MTZ.

^eClinical isolate resistant to FDX.

^fClinical isolate resistant to VAN.

^gStrain HM-709, Gram-negative, anaerobic bacterium that is commensal and critical to host immunity; a minor component of the human gut microflora (<1%).

^hStrain HM-846, anaerobic, Gram-positive bacterium commonly found in the normal human intestinal microflora isolated from human feces, nonsporulating.

ⁱStrain HM-784, Gram-positive, aerobic or facultatively anaerobic bacterium that occurs in the mucosa and normal skin flora of humans and animals.

^jStrain HM-992, anaerobic, nonsporulating, Gram-negative bacterium commonly found in the gastrointestinal tract.

^kStrain HM-102, Gram-positive, anaerobic bacteria commonly found in the normal human gastrointestinal tract, commonly used as a probiotic to maintain the balance of gut microbial flora.

^lStrain HM-644, Gram-positive, facultative, anaerobe bacterium commonly found in the normal human gastrointestinal tract, commonly used in yogurt production as a probiotic to suppress Helicobacter pylori infections.

^mGram-negative, nonsporulating bacterium commonly found in the intestinal tract of humans and animals.

ⁿStrain HM-178, anaerobic, nonsporulating, Gram-positive bacterium commonly found in the gastrointestinal flora of humans and animals.

^oFecal microbiota preparation obtained from OpenBiome tested at a 1000-fold dilution.

Compound 1 is poorly active against MRSA (128-fold lower activity), in contrast to VAN and FDX (Table 1). As broad-spectrum antibiotics can disrupt the normal gut flora allowing for C. difficile to become established, we evaluated the selectivity of compound 1 toward C. difficile ATCC 43255 compared to major gut bacterial flora. Compound 1 was generally more selective (16- to 1000-fold, Table 1) than VAN (0.5- to 64-fold), MTZ (2- to 512-fold), and FDX (0.25- to 512-fold).

The in vitro cytotoxicity of 1 was evaluated by the XTT (2,3-bis(2-methoxy-4-nitro-5-sulfophenyl)-2H-carboxanilide) assay using HeLa cells. The IC₅₀ value of 179 ± 4 μg/mL was >700-fold above its MIC₉₀. As antibiotics for the treatment of CDI need to be poorly absorbed so as to achieve high concentrations in the gut, we determined the levels of compound 1 in plasma and feces after a 20 mg/kg single oral administration to mice. Concentrations of 1 were nonquantifiable in plasma. The levels in feces were 13 μg/g (equivalent to 13 μg/mL, assuming a density of 1 g/mL), >100-fold higher than the MIC for C. difficile ATCC 43255, indicating that desirable high concentrations for effective antibacterial activity are reached in the gut.

Antibacterial activity typically manifests in the logarithmic phase of the bacterial growth and not in the slow-growing cells of the stationary phase of growth. We investigated the effect of compound 1 on stationary-phase C. difficile using time-kill assays. Compound 1 reduced C. difficile stationary-phase growth by 2 log₁₀, whereas VAN decreased it by a mere 1 log₁₀ (Figure 1A). Another attribute of antibiotics is selection for resistance. We investigated resistance development against compound 1 by serial passage evaluation of MIC values and in comparison to VAN. The MIC values for compound 1 increased from 0.125 to 1 μg/mL, whereas those for VAN changed from 0.5 to 4 μg/mL (Figure 1B). In both cases, the increase was 8-fold. However, the MIC values for compound 1 were 4-fold lower.

Figure 1.

Anti-C. difficile activity of compound 1. (A) Compound 1 reduces stationary-phase C. difficile growth better than VAN in the time-kill assay. (B) Serial passage of C. difficile ATCC 43255 in the presence of sub-MIC concentrations of 1 and of VAN.

The mechanism of action of 1 was investigated using macromolecular synthesis assays. These assays measure the incorporation of radioactive precursors ([2,8-³H]-adenine,¹⁶ [5,6-³H]-uridine,¹⁷l-[3,4,5-³H(N)]-leucine,¹⁸ and N-acetyl-d-[6-³H]glucosamine),¹⁹ respectively, into DNA, RNA, protein, and peptidoglycan. As shown in Figure 2A, compound 1 inhibits cell wall peptidoglycan biosynthesis (55 ± 4%). This value was similar to the value (51 ± 1%) for C. difficile exposed to the cell wall-active antibiotic oxacillin, which served as a positive control. Minimal levels of inhibition were observed for DNA, RNA, and protein synthesis. We confirmed that compound 1 targets peptidoglycan biosynthesis by imaging C. difficile bacteria by scanning electron microscopy (Figure 2B–G). This analysis visualizes the cell wall as the outermost layer of the cell envelope of Gram-positive bacteria such as C. difficile. The surface of untreated C. difficile ATCC43255 (Figure 2B) is unperturbed but is altered upon exposure to VAN (Figure 2C). Upon exposure to 1 (Figure 2D shows C. difficile at 4× MIC; Figure 2E–G shows the effects of increased levels of 1) bleb formation on the surface was evident (Figure 2D,E). At higher concentrations of 1, decisive damage to the cell wall (Figure 2F) culminates in the destruction of the bacterium (Figure 2G).

Figure 2.

Mechanism of action of compound 1. (A) Macromolecular biosynthesis assays: Compound 1 was incubated with C. difficile ATCC43255 at 0.5× MIC. The positive controls are ciprofloxacin (DNA, MIC 8 μg/mL), rifaximin (RNA, MIC 0.03 μg/mL), linezolid (protein, MIC 2 μg/mL), and oxacillin (peptidoglycan, MIC 8 μg/mL). The incubation time was 60 min for DNA, 120 min for peptidoglycan synthesis and RNA, and 180 min for protein synthesis. (B–G) Imaging of C. difficile ATCC42355 vegetative cells (logarithmic growth) by scanning electron microscopy. (B) C. difficile ATCC43255 untreated control; (C) VAN-treated at 4× MIC; compound 1 treated at (D) 4× MIC, (E) 8× MIC, (F) 16× MIC, and (G) 32× MIC.

Next, compound 1 was evaluated in a mouse model of recurrent CDI. In this model, vehicle-treated mice die soon after infection, while VAN-treated mice survive initially, with infection resurging on day 10, resulting in a similar survival between vehicle and VAN at the end of the study. This model uses an antibiotic cocktail containing kanamycin, gentamicin, colistin, MTZ, VAN, and clindamycin to perturb the gut microbiota, followed by infection with C. difficile ATCC 43255.²⁰ Oral dosing with compound 1 was once a day for 5 days, with VAN as positive control and vehicle as negative control. This model results in recurrence of infection in VAN-treated mice, as seen by loss of body weight on days 10–15 (Figure 3A). At the end of the study, animals receiving compound 1 showed a similar survival of the mice (Figure 3B) compared to VAN.

Figure 3.

Efficacy of compound 1 in a mouse model of recurrent CDI. (A) Body weights (n = 10 mice per group); recurrence of infection is observed in the VAN-treated group on days 10 to 15. (B) Survival plot, n = 10 mice per group. **p < 0.01 compared to the vehicle by the Student’s t test. Differences in survival were not statistically significant (p > 0.05) by the log rank test.

Notwithstanding the availability of MTZ, VAN, and FDX for the treatment of CDI, the recurrence of infection occurs in 25% of patients treated with these antibiotics.^21,22 Over 12,000 annual deaths from CDI occur in the US alone, leading to the designation of these infections as an urgent health problem. Compound 1 is a potent anti-C. difficile agent with selectivity against this organism.

■ METHODS

Synthesis of Compound 1.

Detailed experimental procedures are given in the Supporting Information.

Antibiotics.

Clindamycin, ciprofloxacin, oxacillin, rifaximin, kanamycin, gentamicin, colistin, vancomycin, and metronidazole were purchased from Sigma-Aldrich (St. Louis, MO); linezolid was purchased from AmplaChem Inc., (Carmel, IN), and fidaxomicin was obtained from BOC Sciences (Shirley, NY).

Bacterial Strains.

A total of 101 C. difficile strains were used in the study. Of the isolates, 86 were from Biodefense and Emerging Infections Research Resources Repository (BEI Resources, Manassas, VA) and eight, from American Type Culture Collection (ATCC, Manassas, VA); seven resistant C. difficile isolates were obtained from Cleveland VA Medical Center. Other gut bacteria strains were obtained from BEI. MRSA strains NRS70 (N315) and NRS119 were obtained through the Network on Antimicrobial Resistance in Staphylococcus aureus (NARSA). ATCC 29213 and E. faecium NCTC7171 (ATCC 19434) were purchased from ATCC (ATCC; Manassas, VA). All the strains were cultured and stored according to the supplier instructions.

Minimum Inhibitory Concentrations (MICs).

MICs for C. difficile strains and the common gut bacteria were determined using broth microdilution techniques as reported earlier using brucella broth supplemented with hemin and vitamin K or supplemented BHIS broth.²³ Lactobacillus MRS broth was used for Lactobacillus strains. Sodium lactate was supplemented in the media for Veillonella sp. The test compounds were added in 2-fold serial dilutions, and the bacteria were added to a final concentration of 5 × 10⁵ CFU mL⁻¹. All incubations, unless specified otherwise, were carried out at 24 or 48 h at 37 °C in an anaerobic chamber (Whitley DG250 workstation, Microbiology International, Frederick, MD). Corynebacterium and Lactobacillus species were incubated aerobically. MIC₅₀ and MIC₉₀ values for compound 1 against 101 C. difficile strains were determined.

XTT Assay.

Cytotoxicity assays were performed in triplicate against HeLa cells (ATCC CCL-2), as previously described.²⁴ The IC₅₀ (the concentration that results in 50% cell viability) values were calculated by nonlinear regression with GraphPad Prism 5 (San Diego, CA).

Pharmacokinetics (PK) Study.

Mice (n = 2 per time point) were given a single 100 μL solution of 5 mg/mL of compound 1 (equivalent to 20 mg/kg). Terminal blood was collected in heparinized syringes by cardiac puncture at 30 min and 1, 2, 4, and 8 h. The two mice at the 8 h time point were place in metabolism cages for the collection of feces. Compound 1 was dissolved in 5% DMSO/95% water with 96 mg/mL of sodium bicarbonate. Blood was centrifuged at 1000g for 20 min at 4 °C, and plasma was separated and stored at −80 °C until analysis. A 45 μL aliquot of plasma was mixed with 5 μL of H₂O/ACN (1:1) and was quenched with 100 μL of acetonitrile containing 20 μM of internal standard (compound 10). The sample was centrifuged at 21 000g for 15 min at 4 °C to precipitate the proteins, and the resulting supernatant was analyzed by the UPLC–TQD detector using multiple-reaction monitoring (MRM). Calibration curves for compound 1 were prepared from 0 to 102.4 μM using a control mouse plasma (50 μL). Peak area ratios relative to the internal standard and linear regression were used from which the concentrations in plasma samples from the PK study were determined.

An average of 250 mg of feces was collected at 8 h from two mice and, subsequently, homogenized with 6 times the feces volume (1500 μL of H₂O/ACN 1:1) using a Bullet Blender (Next Advance, Inc., Troy, NY). The homogenate was centrifuged at 21 000g for 5 min. The fecal supernatant (72 μL) was mixed with 8 μL of a solution of H₂O/ACN (1:1) and 40 μL of 50 μM of the internal standard in ACN with 0.5% formic acid. A calibration curve was prepared from 0 to 102.4 μM in control mouse feces. Plasma and feces were analyzed on a Waters TQD tandem quadrupole detector (Waters Corporation, Milford, MA) using a YMC-Triart C18 100 mm × 2.0 mm column (3 μ, YMC Co. Ltd., Kyoto, Japan). The chromatographic conditions consisted of an elution with 30% H₂O/70% ACN containing 0.1% formic acid for 5 min at a flow rate 0.5 mL/min. Mass spectrometry acquisition was performed in the positive electrospray ionization mode with MRM. The capillary, cone, extractor, and RF lens voltages were set at 2.8 kV, 40 V, and 1 and 0.1 V, respectively. The desolvation and cone gas (nitrogen) flow rates were 650 and 50 L/h, respectively. The source temperature was set at 150 °C. The desolvation temperature was 350 °C. The MRM transitions used were 459.4 → 254.0 for compound 1 and 294.8 → 133.5 for the internal standard.

Time-Kill Assay.

For the time-kill assay, stationary-phase cultures of C. difficile were prepared as described²⁵ and incubated with compound 1 or VAN at 8×, 16×, and 32× MIC in prereduced supplemented brucella broth. A control tube without antibiotic was also included. At intervals of 4, 8, and 24 h, 100 μL aliquots of the cultures were plated onto prereduced brain heart infusion broth agar media for colony counts. The limit of detection was 50 cfu/mL, and a ≥3 log₁₀ reduction of colonies from the starting inoculum was considered bactericidal. The experiment was done in triplicate.

Serial Passage.

A serial-passage assay for the development of resistance in C. difficile strain ATCC43255 was performed in duplicate in prereduced supplemented brucella broth. The starting MIC for the strain against compound 1 and VAN using a broth macrodilution assay was 0.125 and 0.5 μg/mL, respectively. 2-Fold serial dilutions of the drugs were prepared from 2× MIC to 0.25× MIC, and cultures were added to a final concentration of 5 × 10⁵ CFU/mL. Each day, wells with the highest concentration of drugs showing growth were used to inoculate the series of tubes the next day. The process was repeated for 25 passages. After 25 days, cultures showing a 4-fold or higher increase in MIC were confirmed by three passages in drug-free media.

Macromolecular Synthesis Assay.

The macromolecular synthesis assay was performed as described earlier.^16,26 The radioactive precursors and positive controls were [2,8-³H]-adenine and ciprofloxacin for DNA, [5,6-³H]-uridine and rifaximin for RNA, l-[3,4,5-³H(N)]-leucine and linezolid for protein, and N-acetyl-d-[6-³H]glucosamine and oxacillin for the peptidoglycan synthesis. The radiolabeled precursors were purchased from PerkinElmer (Waltham, MA, USA).

Scanning Electron Microscopy.

Bacteria in midexponential growth phase were incubated overnight with 4× to 32× MIC of compound 1. Following incubation, the cells were washed with PBS (3×), applied onto poly-l-lysine (Santa Cruz Biotechnology, Dallas, TX) coated microglass coverslips (Electron Microscopy Sciences, Hatfield, PA), and incubated for 15 min. The cells were fixed for 1 h with 2% glutaraldehyde, followed by a wash (3×) with sodium cacodylate buffer at pH 7.4. The samples were fixed with 1% osmium tetroxide for 1 h and rinsed (3×) in the buffer. The samples were then put through a graded ethanol series for dehydration, followed by critical point drying. The samples were then mounted on SEM stubs and sputter-coated with iridium. Microscopy and imaging were performed on a Magellan 400 XHR instrument (FEI, Hillsboro, OR).

Animals.

Female C57BL/6 mice weighing 18–20 g and 6–8 weeks of age were used for the efficacy study (Jackson, Bar Harbor, ME). Female CD-1 mice (24–31 g and 7–8 weeks of age) were used for the PK study. They were housed in groups of five mice in sterile polycarbonate shoeboxes with bedding consisting of 1 in. corncob (The Andersons Ind., Maumee, OH) and Alpha-Dri (Shepherd Specialty Papers, Inc., Richland, MI). Mice were maintained on a 12 h light/dark cycle at 72 °F and were given a Teklad 2918 irradiated extruded rodent diet and water ad libitum. All procedures were performed in accordance with and approved by the University of Notre Dame Institutional Animal Care and Use Committee.

C. difficile Infection Model.

The recurrent C. difficile infection model reported by Chen et al.²⁰ was used. Mice (n = 10 per group, total of 40 mice per study) were given an antibiotic cocktail containing kanamycin (0.4 mg L⁻¹), gentamicin (0.035 mg L⁻¹), colistin (850 U mL⁻¹), metronidazole (0.215 mg L⁻¹), and vancomycin (0.045 mg L⁻¹) in sterile drinking water containing 5% sucrose for 3 days, followed by 2 days of regular water. Subsequently, mice were given a 10 mg/kg intraperitoneal injection of clindamycin. The following day, the mice were infected intragastrically with 10⁴ CFU of C. difficile ATCC 43255. Mice were given single oral doses of compound 1 at 20 and 40 mg/kg, VAN at 50 mg/kg, or vehicle (5% DMSO/95% water) once a day for 5 days. Body weights and survival were recorded for 25 days. The study was repeated a second time. Body weights and survival curves were analyzed for statistical significance using the Student’s t test and log rank (Mantel-Cox) test in GraphPad Prism. Differences in survival were not statistically significant (p > 0.05).