Reutericyclin and related analogues kill stationary phase Clostridium difficile at achievable colonic concentrations

Abstract

Objectives

The stationary phase of Clostridium difficile, which is associated with the symptoms of the diarrhoeal disease, is refractory to antibiotic killing. The aim of this study was to explore whether probiotic-derived reutericyclin and related synthetic analogues could kill stationary phase C. difficile at concentrations achievable in the gastrointestinal tract.

Methods

The bactericidal activities of reutericyclin and lead compound derivatives were examined against logarithmic and stationary phase cultures of different C. difficile strains. The absorption of compounds across the intestinal epithelia was tested using the Caco-2 permeability model.

Results

Unlike vancomycin and metronidazole, reutericyclins demonstrated concentration-dependent killing, being rapidly bactericidal against both logarithmic and stationary phase cells, at low concentrations (0.09–2 mg/L). The intestinal absorption of unmodified reutericyclin was poor and comparable to that of vancomycin. However, this property varied significantly for the synthetic reutericyclin analogues, ranging from well absorbed to non-absorbed. The non-absorbable compounds were highly effluxed, suggesting this parameter could be modulated to obtain agents with superior efficacy.

Conclusions

Reutericyclins showed excellent potency against the lethal non-growing stage of C. difficile at concentrations that may be attained in the gastrointestinal tract. Since these agents represent novel potential treatments for C. difficile infection, further development of this compound class is warranted.

Introduction

Clostridium difficile infection (CDI) is the leading cause of antibiotic-associated diarrhoea in hospitalized patients.¹ Recently the incidence and severity of CDI has substantially increased, particularly among elderly patients. This is due to the emergence of hypervirulent strains identified as BI/NAP1/027 (NAP1) that are reported to demonstrate robust production of the lethal enterotoxins A and B.² C. difficile produces these toxins in its stationary phase and this physiological state is likely to present a challenge to effective chemotherapy, since slow and non-growing cells are often refractory to antibiotic-mediated killing.³ For more than two decades there has been almost exclusive reliance on oral metronidazole or vancomycin for CDI, but therapeutic failures are now common.¹ It is therefore evident that new therapies with low oral bioavailability, achieving high local concentrations at the site of infection, are needed to combat CDI. Importantly, if these agents also rapidly kill stationary phase cells, this could potentially mitigate the effects of toxin production by removing the offending pathogen.

The naturally occurring tetramic acid reutericyclin, produced by strains of Lactobacillus reuteri, is a small non-peptide antibiotic that selectively dissipates the bacterial transmembrane potential, resulting in narrow-spectrum activity against Gram-positive bacteria.⁴ Recently, we expanded reutericyclin as an emerging class of antibiotics through the synthesis of several analogues that show improved antibacterial properties.^4,5 The activity of reutericyclins against C. difficile has not been described. Therefore, we explored whether these agents (Table 1) possess desirable in vitro properties warranting consideration for development as anti-C. difficile therapies.

Table 1.

Anti-C. difficile activity and permeability across the intestinal epithelia by reutericyclins and comparator agents

	Anti-C. difficile and absorption properties
	Reutericyclin	867	1138	Vancomycin	Metronidazole
Chemical structurea
Anti-C. difficile activity (mg/L)b,c
MIC vs strain 9689	0.19 (0.12–0.25)	0.09 (0.06–0.12)	0.25	0.19 (0.12–0.25)	0.19 (0.12–0.25)
MBClog vs strain 9689	0.38 (0.25–0.5)	0.19 (0.12–0.25)	0.5	1	0.38 (0.25–0.5)
MBCSTA vs strain 9689	0.5	0.75 (0.5–1)	0.75 (0.5–1)	>64	24 (16–32)
MIC vs strain 1803	0.09 (0.06–0.12)	0.12	0.5	1	0.19 (0.12–0.25)
MBClog vs strain 1803	0.25	0.25	2	2	0.38 (0.25–0.5)
MBCSTA vs strain 1803	0.5	0.5	4	>64	48 (32–64)
MIC vs strain 1875	0.09 (0.06–0.12)	0.09 (0.06–0.12)	0.5	0.25	0.09 (0.06–0.12)
MBClog vs strain 1875	0.19 (0.12–0.25)	0.19 (0.12–0.25)	1	0.38 (0.25–0.5)	0.31
MBCSTA vs strain 1875	0.38 (0.25–0.5)	0.38 (0.25–0.5)	2	>64	36 (8–64)
Permeability across colonic Caco-2d,e
Papp A to B (nm/s)	133.1 ± 42.1	6.8 ± 3.4	481.7 ± 72.2	98.1 ± 17.1	574 ± 55.6
Papp B to A (nm/s)	93.6 ± 4.4	128.2 ± 34.7	345.8 ± 24.1	55.0 ± 6.3	330.6 ± 27.6
Efflux ratio (B → A/A → B)	0.7	18.9	0.7	0.6	0.6

^aThe numbers of additional carbons making up the aliphatic chain are shown by the number adjacent to the parentheses in the shown structures; the only difference between 867 and 1135 (not shown) is that the latter carries an isopropyl group at the 5-position instead of the isobutyl shown in 867, whereas 1141 (not shown) is a stereoisomer of 867.

^bAll MBCs are the average of at least two independent tests; the activity range is shown in parentheses.

^cMBCs against logarithmic and stationary phase cells are shown as MBC_log and MBC_STA, respectively; MBCs of reutericyclins are ∼2- to 8-fold their respective MICs.

^dData are based on three replicates; by convention a ratio of >2 specifies that the compound is a substrate of an efflux pump, probably P-glycoprotein.

^e1135 (Papp A to B = 19.4 ± 6.7; Papp B to A = 181.6 ± 56.3; efflux ratio = 9.6); 1141 (Papp A to B = 7.8 ± 3.4; Papp B to A = 221.9 ± 129.8; efflux ratio = 28.6); carbamazepine, control for high permeability (Papp A to B = 646.9 ± 80.5; Papp B to A = 312.4 ± 23.1; efflux ratio = 0.5); digoxin, control for low permeability and P-glycoprotein efflux (Papp A to B = 54.3 ± 6.2; Papp B to A = 142.8 ± 6.2; efflux ratio = 2.6).

Materials and methods

Antibiotics, bacterial strains and growth

Reutericyclin and its analogues 867, 1135, 1138 and 1141 were synthesized as previously described;⁴ metronidazole and vancomycin were from Sigma-Aldrich (St Louis, MO, USA). C. difficile 9689 (toxinotype 0), BAA-1803 (toxinotype III, NAP1) and BAA-1875 (toxinotype V, NAP7) were obtained from ATCC. Unless otherwise stated, all strains were grown in pre-reduced TY broth or Brucella agar at 37°C in an Anoxomat Mart II system (Mart Microbiology B.V., Drachten, The Netherlands).

Determination of MBCs

MBCs were determined against logarithmic (MBC_log) and stationary phase (MBC_STA) cultures. The MBC_log was determined against logarithmic cells (10⁶ cfu/mL, from culture of OD₆₀₀ ∼ 0.5) in 24-well microtitre plates (Costar; Corning Incorporated, Corning, NY, USA). After 24 h the MIC was recorded as the lowest concentration of antibiotic that prevented visible growth. Subsequently the MBC_log was defined as the lowest concentration of antibiotic killing ≥3 logs of cells. MBC_STA was similarly determined, but used stationary phase cultures (24 h, OD₆₀₀ ≥ 1) containing 10⁸–10⁹ cfu/mL. All MBCs were performed at least twice. Plating was done on Brucella agar to avoid the effect of carryover antibiotic, since serum content reduces the activity of reutericyclins.⁵

Time–kill studies

After exposing stationary phase cultures to antibiotics or drug-free controls, viable counts and OD₆₀₀ readings were performed at 2, 4, 6 and 24 h. Experiments were performed on two independent replicates.

Cytotoxicity

This was performed against colonic carcinoma Caco-2 cells exactly as previously described.⁵

Caco-2 drug permeability assay

The permeability of compounds across the colonic epithelia was evaluated using functional Caco-2 cells exactly as previously described.⁶

Results

Reutericyclins are bactericidal against log-phase C. difficile

As shown in Table 1, reutericyclin and the main lead compounds 867 and 1138 were highly active against three unrelated test strains, with mean MICs of 0.09–0.5 mg/L. Significantly, the agents were bactericidal, causing a >3 log reduction in cells, at concentrations close to their respective MICs (i.e. ≤4-fold), and their activities were comparable to those of metronidazole and vancomycin (Table 1).

Bactericidal activity is retained against stationary phase C. difficile

Interestingly, reutericyclins retained potent bactericidal activity against all 24 h stationary phase cultures, with killing at concentrations that were ∼2- to 8-fold above their respective MICs (Table 1). This unusual finding for the killing of stationary phase C. difficile by reutericyclins is a property not exhibited by many established classes of antibiotics against several pathogens.³ Indeed, both vancomycin and metronidazole lost their activities against stationary phase C. difficile.

Rapid concentration-dependent killing of stationary phase cultures

Based on time–kill kinetic studies, reutericyclins displayed rapid and concentration-dependent killing (Figure 1). Within 2 h compound 867 at 16 mg/L reduced the viability of the NAP1 strain BAA-1803 by an average of 3 logs (Figure 1a), reaching the limit of detection (i.e. 10² cfu/mL) in 6 h. At a 4-fold lower concentration of 4 mg/L, viability was reduced by ∼2.9 log in 2 h, but required more than 6 h of exposure to reach the limit of detection. Killing in excess of 3 logs was reached in 4 h at a low concentration of 1 mg/L, while at 0.25 mg/L a maximum kill of only 2.3 logs was achieved in 24 h. Similar findings were also observed against strain 9689 (Figure 1b), supporting the conclusion that reutericyclins exert rapid concentration-dependent killing. The killing of stationary phase cells did not result from cell lysis as determined from OD₆₀₀ measurements (data not shown), in agreement with previous findings in staphylococci.⁵ In spite of their excellent antibacterial activities against vegetative cells, reutericyclins, like most other antibiotics, including metronidazole and vancomycin, were inactive against purified spores (data not shown).

Figure 1.

Time–kill kinetics against stationary phase cultures of (a) an NAP1 epidemic strain, BAA-1803 and (b) ATCC 9689. The lead compound 867 is compared with metronidazole (MTZ) and vancomycin (VAN) at concentrations shown in the legend. Each point represents the mean of duplicate determinations. The limit of detection is 100 cfu/mL.

Bactericidal concentrations are achievable in gastrointestinal tract

Using the Caco-2 permeability model, some reutericyclins were found to be non-absorbed (Table 1). Such agents may achieve high local concentrations in the gastrointestinal tract and may permit rapid killing of C. difficile. In this model, colonic Caco-2 cells are grown as monolayers that form tight junctions and display functional properties of intestinal epithelial cells. The apical to basolateral side represents the normal orientation of intestinal epithelia in the gastrointestinal tract. Therefore, corresponding permeability coefficients (Papp A/B) indicated that metronidazole (574 ± 55.6 nm/s) was the most permeable compound, consistent with previous reports that it is highly absorbed.⁷ In contrast, oral vancomycin is poorly absorbed from the gastrointestinal tract,⁷ which was evident from the Caco-2 assay, where the permeability to vancomycin was much lower than metronidazole (98.1 ± 17.1 nm/s). Interestingly, the permeability to unmodified reutericyclin (133 ± 42.1 nm/s) was comparable to that of vancomycin. However, replacement of its chemically unstable α,β-unsaturated group with either the bi-cyclic ring in 1138 or straight-chain alkyl substituent in 867 had contrasting effects. While 1138 (481.7 ± 72.2 nm/s) showed substantial permeability, 867 (6.8 ± 3.4 nm/s) was relatively non-absorbed. To examine if non-permeable 867 was subject to asymmetric transport, the Papp from the basolateral to the apical side was measured. Consequently 867 (128.1 ± 34.7 nm/s) displayed high Papp B/A transfer with a high efflux ratio of 28.6. Hence non-absorption of 867 likely results from its asymmetric transport by an efflux pump. In support of these findings, the structurally similar analogues of 867 (Table 1), 1135 (MIC 0.25 mg/L against 9689) and 1141 (MIC 0.12 mg/L against 9689) were also highly effluxed and non-absorbed (Table 1). Owing to their lipophilicity, all reutericyclins were permeable in the parallel artificial membrane permeability assay (PAMPA; data not shown), which only examines the physicochemical properties for permeation and does not evaluate biological mechanisms such as transport that influences permeation.

It was revealed that reutericyclins are also not cytotoxic to the intestinal epithelium, since the compounds IC₅₀s were higher than the maximum test concentration (at 200 mg/L >80% of Caco-2 cells were viable).

Discussion

The need for novel anti-infectives to control diarrhoeal infections caused by C. difficile is vital. Currently several approaches such as toxin-binding polymers, passive antibodies, probiotics and new antimicrobials are undergoing clinical evaluation.¹ Because antimicrobials produced by probiotic organisms may be part of a natural mechanism of eliminating C. difficile, we investigated the premise that the probiotic-derived tetramic acid antibiotic reutericyclin could be a novel approach for controlling C. difficile. Reutericyclin, unlike bacteriocins, is resistant to enzymatic proteolysis,⁸ easy to synthesize and easy to chemically modify in order to improve its antibacterial and physicochemical properties.⁴

Our results indicate that reutericyclins exert rapid bactericidal activity against non-dividing stationary phase cells at concentrations close to those causing growth inhibition of logarithmic cells. Therefore, their anti-C. difficile actions are independent of C. difficile growth. Previously, reutericyclins were reported to exhibit bactericidal properties against log-phase and biofilm cells of staphylococci, but this occurred at higher concentrations (e.g. ≥25 mg/L for 867)⁵ than those described herein for the killing of C. difficile. This would suggest that C. difficile is more sensitive to dissipation of its transmembrane potential by reutericyclins. Although the effect of these agents on toxin production was not measured, it is conceivable that an overall decrease in toxin levels would result from the rapid elimination of viable C. difficile. This concept, however, does not apply to vancomycin and metronidazole, which are inactive against stationary phase cells and have little or no effect on toxin production.⁹

The concentration-dependent killing of reutericyclins coupled with the ability for some analogues to be non-absorbed is potentially advantageous for treating CDI. The non-absorption of 867 and related compounds probably resulted from the activity of intestinal efflux pumps, such as P-glycoprotein. However, the permeability of 1138 raises the possibility that the N-substituent position acts as a site that could be modulated to obtain compounds with some permeability for the killing of intracellular C. difficile.¹⁰

Recently Hurdle et al.³ stated that antibiotics specifically targeting the bacterial membrane may be advantageous in treating recalcitrant infections, since the membrane is essential in both dividing and non-dividing cells. The activities of the membrane-active antibiotics daptomycin, oritavancin and tetramic acid derivatives of N-acylhomoserine lactones have been reported against C. difficile, but their effects on its stationary phase cells are unknown. Further studies with reutericyclins and other membrane-active molecules are therefore needed to determine the molecular basis for cell killing and if these agents constitute a new therapeutic paradigm for rapidly resolving CDI. Studies to expand the anti-C. difficile structure–activity relationship and evaluate the spectrum of activity, pharmacokinetics, efficacy and effects on toxin production are warranted to advance reutericyclins as potential therapies for CDI.

Funding

Funding for this research was provided by National Institutes of Health grant ARRA R01AI079653, Cancer Center core grant CA21765 and the American Lebanese Syrian Associated Charities (ALSAC).

Transparency declarations

None to declare.