Synthetic oxepanoprolinamide iboxamycin is highly active against human pathogen Listeria monocytogenes

Listeriosis is a dangerous food-borne bacterial disease caused by the Gram-positive Bacillota (Firmicute) bacterium Listeria monocytogenes. In this report, we show that the synthetic lincosamide iboxamycin is highly active against L. monocytogenes and can overcome the intrinsic lincosamide resistance mediated by VgaL/Lmo0919, a member of ABCF ATPase resistance determinants that act by directly removing the antibiotic from the ribosome. While iboxamycin is not bactericidal against L. monocytogenes, it displays a pronounced postantibiotic effect, which is a valuable pharmacokinetic feature. Experiments in L. monocytogenes infection models are necessary to further assess the potential of iboxamycin as a novel drug for treatment of listeriosis. We demonstrate that VmlR ARE ABCF of Bacillota bacterium Bacillus subtilis grants significant (33-fold increase in MIC) protection from iboxamycin, while LsaA ABCF of Enterococcus faecalis grants an 8-fold protective effect. Furthermore, the VmlR-mediated iboxamycin resistance is cooperative with that mediated by the Cfr 23S rRNA methyltransferase resistance determinant, resulting in up to a 512-fold increase in MIC. Therefore, emergence and spread of ABCF ARE variants capable of defeating next-generation lincosamides in the clinic is possible and should be closely monitored.

In this report, using lincomycin and clindamycin as reference compounds, we i) characterised the efficacy of iboxamycin against L. monocytogenes, ii) probed its ability to specifically counter resistance mediated by ABCFs L. monocytogenes Lmo0919, E. faecalis LsaA and B. subtilis VmlR, iii) characterised its bactericidal/bacteriostatic mechanism of action and, finally, iv) assessed the strength of its post-antibiotic effect (PAE).

Results
L. monocytogenes is highly sensitive to iboxamycin despite VgaL/Lmo0919 ABCF resistance factor To test the lincosamide sensitivity of L. monocytogenes we used two widely-used wild-type strains, both belonging to serovar 1/2a: EGD-e [40] and 10403S, a streptomycin-resistant variant of 10403 [41]. The two wild types are genomically distinct, e.g. the virulence master-regulator PrfA is overexpressed in EGD-e and the prophage content differs between the two strains [42]. In additional to the two wild types, we also tested a L. monocytogenes EDG-e derivative that was genomically modified to abrogate the expression of VgaL/Lmo0919 PLSA resistance factor (EDG-e Δlmo0919) [23].
Both wild-type L. monocytogenes strains are dramatically more sensitive to iboxamycin (MIC of 0.125-0.5 µg/mL) as compared to clindamycin (MIC of 1 µg/mL) and lincomycin (MIC of 2-8 µg/mL) ( Table 1). In agreement with the higher sensitivity of Δlmo0919 EDG-e to lincomycin [23], this strain is 2-8-fold more sensitive to iboxamycin than the corresponding wild type. This indicates that while VgaL does confer some protection from iboxamycin, the high potency of the synthetic antibiotic would likely allow the drug to overcome resistance in clinical settings. A likely explanation is that increased affinity of the synthetic drug for the ribosome renders antibiotic displacement by ABCF ATPases inefficient.
Importantly, expression of Lmo0919 is not constitutive: it is elicited by antibiotic-induced ribosomal stalling on the regulatory short open reading frame upstream of the lmo0919 gene [38]. Therefore, the difference in iboxamycin sensitivity between wild-type and Δlmo0919 EDG-e strains reflects both the ability of Lmo0919 to protect the ribosome from the antibiotic as well as the efficiency of iboxamycin-mediated induction of Lmo0919. To deconvolute these two effects, we used engineered strains that allow for ectopic inducible expression of ABCF in the following experiments.

E. faecalis ABCF LsaA grants a moderate protective effect against iboxamycin
To test the ability of other ABCF PLSA resistance factors to confer resistance to iboxamycin, we compared a pair of E. faecalis strains: one lacking the chromosomally-encoded LsaA (DlsaA pCIEspec) and the other allowing cCF10-peptide-inducible expression of LsaA (DlsaA pCIEspec LsaA) [23]. Using this experimental set up, we could specifically assess the ability of LsaA to protect the strain from lincosamides. While expression of LsaA dramatically increases resistance to clindamycin and lincomycin (96-to 256-fold, respectively), it results in a mere 8-fold protective effect against iboxamycin (MIC of 0.0625 and 0.5 µg/mL, respectively) ( Table 1), demonstrating that iboxamycin can also largely overcome LsaA-mediated resistance.
B. subtilis ABCF VmlR acts cooperatively with rRNA methyltransferase Cfr to grant significant protection against iboxamycin Next we tested a set of B. subtilis strains: wild-type 168 B. subtilis, DvmlR (VHB5) as well as a DvmlR strain in which VmlR is expressed under the control of IPTG-inducible Phy-spank promotor (VHB44) [43] ( Table 1). Disruption of vmlR results in a 33-fold increase in iboxamycin sensitivity (MIC of 2 and 0.06 µg/mL, respectively), and resistance is restored upon ectopic expression of VmlR (MIC of 4 µg/mL, 2fold increase over the wild-type levels). The iboxamycin sensitivity of Δlmo0919 L. monocytogenes EDG-e and DvmlR B. subtilis is near-identical, indicating that the 16-/4-fold difference in iboxamycin sensitivity between wild-type L. monocytogenes and B. subtilis is due to the different efficiency of resistance granted by Lmo0919 and VmlR respectively. Importantly, VmlR loss results in the same relative increase in sensitivity to all lincosamides tested -iboxamycin, clindamycin and lincomycin; 32-33-fold -regardless of the potency of the lincosamide ( Table 1). This suggests that if the affinity of iboxamycin to the target were to be decreased by, for instance, rRNA modification, direct target protection by the ABCF could cooperatively lead to high levels of resistance. To probe this hypothesis, we have characterised the lincosamide sensitivity of B. subtilis strains that express Cfr 23S rRNA methyltransferase under the control of IPTG-inducible Phy-spank promotor, either in the presence or absence of the chromosomally-encoded VmlR. Ectopic expression of Cfr in vmlR+ B. subtilis effected a cooperative resistance to iboxamycin, resulting in MICs of 16-32 µg/mL as opposed to 2 µg/mL when either of these resistance determinants are expressed individually ( Table 1). As expected, Cfr also granted high levels of lincomycin and clindamycin resistance when ectopically expressed in both wild-type and DvmlR strains (MIC ranging from 320 to excess of 640 µg/mL).
Iboxamycin is bacteriostatic against L. monocytogenes and displays a strong postantibiotic effect Macrolide antibiotics that tightly bind the ribosome and dissociate slowly are bactericidal, while macrolides that dissociate rapidly are bacteriostatic [44]. As with lincomycin and clindamycin, iboxamycin was shown to be bacteriostatic against a panel of bacterial species [14]. However, since effects on L. monocytogenes were not assessed in the original report -and the species is highly sensitive to iboxamycin -we tested for potential bactericidal effects of iboxamycin against this pathogen. The three L. monocytogenes strains that we used for the MIC measurements -wild-type 10403 and EGD-e as well as ABCF-deficient EDG-e Δlmo0919were treated with 4x MIC concentration of either iboxamycin, clindamycin and lincomycin for increasing periods of time (from 2 to 24 hours), washed, and then plated on BHI agar plates that contain no antibiotic. The bacterial growth expressed in Colony Forming Units, CFU, was scored after either 24-or 48-hour incubation of plates at 37 °C. When the colony counting was performed after 24 h, we observed potentially bactericidal behaviour of iboxamycin, with almost a two log10 drop in CFU after the 10-hour treatment with the antibiotic (Figure 2A-C). Importantly, no similar CFU decrease was observed for either clindamycin or lincomycin (Figure 2A-C). However, this apparent CFU drop effect of iboxamycin disappeared after 48 h of incubation ( Figure 2D-F), suggesting slow regrowth rather than cidality and indicative of the so-called postantibiotic effect (PAE) [45,46]. PAE is characterised by the time after antibiotic removal where no growth of the treated bacteria is observed. This prolonged action of iboxamycin has been previously noted for S. aureus and E. faecium [14]. Therefore, we next performed post-antibiotic effect experiments in L. monocytogenes, demonstrating that, indeed, iboxamycin displays pronounced PAE, suppressing the growth of the wildtype 10403S and wild-type EGD-e for 6 and 8 hours, respectively ( Figure 3B,C). Clindamycin demonstrates a weaker PAE against EGD-e (2 hours) and similar PAE against 10403S. No clear PAE is detectible for lincomycin. Compared with the isogenic wild-type, EDG-e Δlmo0919 displays similar PAE in the case of clindamycin, and, possibly, somewhat more pronounced PAE in the case of iboxamycin.

Discussion
In this report we have evaluated the efficiency of the oxepanoprolinamide iboxamycin against L. monocytogenes. The antibiotic can largely overcome the intrinsic PLSA resistance of this species that is mediated by the ribosome-associated ATPase VgaL/Lmo0919, and can similarly counteract the intrinsic resistance mediated by ARE ABCF LsaA in E. faecalis. ARE ABCF PLSA resistance factors are broadly distributed among bacterial pathogens [20,22,47,48], and therefore the ability of iboxamycin to largely counteract the ABCF-mediated resistance is a valuable feature of the new antibiotic. However, given that B. subtilis VmlR does confer significant levels of iboxamycin resistance (33-fold increase in MIC) and is cooperative with the Cfr rRNA methyltransferase resistance determinant, emergence and spread of ABCF ARE variants capable of defeating next-generation lincosamides in the clinic is possible and should be closely monitored.
Furthermore, we demonstrate that iboxamycin displays a strong PAE against L. monocytogenes, compromising bacterial re-growth for many hours post antibiotic removal. The PAE is considerably stronger than that of clindamycin while lincomycin displays no PAE. It is possible that the strength of the PAE reflects how tightly the antibiotic binds to the target, the ribosome -and how slowly it dissociates from it. The pronounced PAE suggests that development of even more tightbinding lincosamides could produce effectively bactericidal drugs in the context of infection. Further biochemical studies are necessary to substantiate this hypothesis. Experiments in L. monocytogenes infection models are necessary to further assess the potential of iboxamycin as a novel drug for the treatment of listeriosis.

Synthesis of Iboxamycin
Iboxamycin was prepared according to the method reported by Mason et al. [49].
Liquid growth assays L. monocytogenes was pre-grown on BHI agar plates at 37 °C for 48 hours. Individual fresh colonies were used to inoculate 2 mL of MH-F broth in 15 mL round bottom tubes, which were then incubated overnight at 37 °C with shaking at 180 rpm. The overnight cultures were diluted then with MH-F broth to final OD600 of 0.005 and incubated for 8 hours in a water bath shaker (Eppendorf™ Inova™ 3100 High-Temperature) at 37 °C with shaking at 160 rpm. bacterial growth was monitored by OD600 measurements every 30 minutes.
L. monocytogenes strains were grown in MH-F broth inoculated with 5 x 10 5 CFU/mL (OD600 of approximately 0.0015) with increasing concentrations of antibiotics. After 24-48 hours of incubation at 37 °C without shaking, the presence or absence of bacterial growth was scored by eye.
E. faecalis strains were grown in BHI media supplemented with 2 mg/mL kanamycin (to prevent lsa revertants), 0.1 mg/mL spectinomycin (to maintain the pCIEspec plasmid), 100 ng/mL of cCF10 peptide (to induce expression of LsaA protein) as well as increasing concentrations of antibiotics, was inoculated with 5 × 10 5 CFU/mL (OD600 of approximately 0.0005) of E. faecalis DlsaA (lsa::Kan) strain TX5332 transformed either with empty pCIEspec plasmid, or with pCIEspec encoding LsaA. After 16-20 hours at 37 °C without shaking, the presence or absence of bacterial growth was scored by eye.
B. subtilis strains were grown in LB medium supplemented with increasing concentrations of antibiotics was inoculated with 5 x 10 5 CFU/mL (OD600 of approximately 0.0005), and after 16-20 hours at 37 °C without shaking the presence or absence of bacterial growth was scored by eye.

Time-kill kinetics assay
The protocol was based on that of [51] and Svetov [44]. Exponential L. monocytogenes cultures in MH-F broth (OD600 ≈ 0.3) were diluted to 10 5 CFU/mL (OD600 = 0.001) in 10 mL of MH-F broth either supplemented with appropriate antibiotic at four-fold MIC concentration or without antibiotics (positive growth control), and the resultant cultures were incubated at 37 °C without shaking. 1 mL aliquots were taken at incramental incubation times (0, 2, 4, 6, 8 and 10 h), spun down at 4000 rpm for 5 min at room temperature and cell pellets were gently washed twice with 900 µL of 1x PBS. Cell pellets were resuspended in 100 µL of 1x PBS, ten-fold serial dilutions were prepared in 96-well plates (10 -1 -10 -8 ), and 10 µL resultant ten-fold seral dilutions were spotted on BHI agar plates. Colony forming units were scored after 24-to 48-hour incubation at 37 °C.

Post Antibiotic Effect (PAE) assay
Exponential cultures of L. monocytogenes strains in MH-F blood broth media (OD600 ≈ 0.3) were diluted to 10 5 CFU/mL (≈OD600 of 0.001) in 5 mL of MH-F media either supplemented with appropriate antibiotic at four-fold MIC concentration or without antibiotics (positive growth control) and incubated at 37 °C for without shaking for 2 h. After the 2 h pre-treatment, antibiotics were removed by 1:100 dilution of 100 µL into 10 mL of fresh prewarmed MH-F blood broth media. At incremental time points (0, 2, 4, 6, 8 and 10 h), 1 mL of the 100x-diluted cell culture was harvested, centrifuged for 5 min at 4000 rpm, 900 µL of the medium was removed, and the pellets were resuspended in the remaining 100 µL. The volume was adjusted to 1 mL with 1x PBS. Control cultures without antibiotics were handled similarly. Cell solutions were then serially diluted ten-fold to 10 -8 , and 10 µL were spotted on BHI agar plates. Plates for individual time points were incubated at room temperature until the last set of plates were spotted (10 h time point), and then incubated at 37 °C. The plates were scored after 24 and 48 h incubation at 37 °C and imaged using ImageQuant LAS 4000 (GE Healthcare). The last time point (24 h) was processed separately analogously to 0-10h time points (see above).
(VH). KJYW was supported by a National Science Scholarship (PhD) by the Agency for Science, Technology and Research, Singapore.

Conflicts of Interest
AGM is an inventor in a provisional patent application submitted by the President and Fellows of Harvard College covering oxepanoprolinamide antibiotics described in this work. AGM has filed the following international patent applications: WO/2019/032936 'Lincosamide Antibiotics and Uses Thereof' and WO/2019/032956 'Lincosamide Antibiotics and Uses Thereof'.

Table 1. Broth microdilution Minimum inhibitory concentration (MIC) testing of lincosamide antibiotics against L. monocytogenes, E. faecalis and B. subtilis strains.
In the case of L. monocytogenes strains, MIC testing was carried out in MH-F broth and growth inhibition was scored after 48 hours incubation at 37 °C. E. faecalis MIC testing was carried out in BHI broth supplemented with 2 mg/mL kanamycin (to prevent lsa revertants), 0.1 mg/mL spectinomycin (to maintain the pCIEspec plasmid), 100 ng/mL of cCF10 peptide (to induce expression of LsaA protein). B. subtilis MIC testing was carried out in either LB medium or LB supplemented with 1 mM IPTG to induce expression of either VmlR or Cfr protein, and growth inhibition was scored after [16][17][18][19][20]    To determine the time taken for antibiotic treated L. monocytogenes strains to resume growth after a two-hour antibiotic treatment, exponentially growing type strains; 10403S (A), EDG-e (B) or VgaAdeficient EDG-e Δlmo0919 (C) were treated with 4x MIC of either iboxamycin, clindamycin, lincomycin, or no antibiotic as control, for two hours. Cells were then diluted by 100-fold to remove the antibiotic, and samples taken every two hours subsequently for viability counting. All experiments were carried out in MH-F broth at 37 °C with shaking at 180 rpm, data points are from three biological replicates and standard deviation is indicated with error bars.