Antibiotic–Adjuvants Abolish Resistance Conferred by the Staphylococcus aureus Erythromycin Resistance Methyltransferase in an Escherichia coli Model

Abstract

Enzyme-mediated resistance is among the main strategies that bacteria use to evade antibiotic action. S-Adenosylmethionine-dependent erythromycin resistance methyltransferases catalyze the methylation of 23S rRNA in bacteria, causing resistance to macrolides, lincosamides, and streptogramin type-B antibiotics. Given the diversity and number of identified variants of these enzymes, it is vital to devise ways of inhibiting their activity to rescue affected antibiotics. Here, we use computer-aided solvent mapping and virtual screening techniques to identify inhibitors of Erms displaying promising adjuvant properties. We further demonstrate that an E. coli model expressing a recombinant S. aureus ErmC (SaErmC) variant causes substantial resistance to representative macrolide and lincosamide antibiotics. Assessment of test compounds using this resistance model revealed candidates that displayed promising adjuvant activity when combined with erythromycin or clindamycin. Antibiotic combinations with a principal candidate oxadiazole, JNAL-016, completely blocked SaErmC-mediated resistance against erythromycin, resulting in an antibiotic-sensitive phenotype in broth microdilution screening assays. This compound also suppressed ErmC activity, allowing erythromycin to regain its bactericidal properties when assessed in actively growing cultures using time-kill assays. JNAL-016 displayed a noncompetitive mode of inhibition against SaErmC activity in vitro and bound the purified enzyme with high affinity (K _d = 1.8 ± 0.7 μM) based on microscale thermophoresis data. Competition experiments suggested that JNAL-016 competes with SAM for its binding pocket on the enzyme, and this compound exhibited no toxicity against human embryonic kidney cells. These findings establish a practical strategy for targeting Erm-mediated resistance, which could lead to a viable adjuvant-based therapy against bacterial pathogens that weaponize variants of this class of methyltransferases.

The prevalence of infections caused by antibiotic-resistant bacterial pathogens remains a significant health challenge that requires urgent attention, especially given the sharp decline of new antibiotic candidates emerging from the FDA approval pipeline over the past few decades. , Current statistics estimate about 2.8 million infections and 35,000 deaths occurring annually from resistant microbes in the U.S., but these numbers are more alarming from a global perspective, with dire projections by 2050 if the problem is left unresolved. , Bacterial evolution, including the rampant spread of old and newly acquired resistance mediators, makes tackling this issue daunting. Confronting this task is particularly challenging when considering species with acquired multidrug resistance mechanisms, commonly referred to as superbugs, which often lead to untreatable life-threatening clinical infections. , Such microorganisms use myriad strategies to survive antibiotic action, including reduced uptake or efflux of drugs from the cell, and target or antibiotic-modifying enzymes. , Given the diversity and complexity of these bacterial defense tactics, it is vital to devise countermeasures for targeting such weaponry to circumvent their survival mechanisms and avoid a looming health catastrophe.

Enzyme-mediated resistance is among the primary strategies pathogenic bacteria use to counteract antibiotic action. , These enzymes are mainly located on transmissible mobile genetic elements, like plasmids, which facilitate their rapid interspecies spread through horizontal gene transfer, increasing the health risk level to animals and humans considerably. − Resistance-conferring enzymes are typically not essential for bacterial-cell viability, and several classes are currently clinically active, neutralizing the effectiveness of many antibiotics, including those designated as our last line of defense. , The nonessential character and prevalence of such enzymes conveniently provide a potential therapeutic strategy for developing antibiotic-adjuvant-based treatment regimens. These treatment options would evade associated resistance mechanisms linked to the affected classes of antibiotics by neutralizing the activity of the enzyme(s) responsible. , Ultimately, this approach would restore the activity of doomed antibiotics that succumbed to similar bacterial defenses, allowing for precious time for the discovery and development of more efficacious antibiotics to fight resistant pathogens. Also, targeting nonessential pathways would avoid any potential selective pressure, eliminating the requirement for a swift adaptive survival response by bacterial cells. Several prominent, well-documented classes of enzymes facilitate widespread antibiotic resistance, including β-lactamases and aminoglycoside-modifying enzymes. − Research over the past few decades on the former class provides a glimpse of the potential effectiveness and application of adjuvant-based options for tackling antibiotic resistance mediated by enzymes. ,

Bacterial methyltransferases are among the known resistance-conferring enzymes. Despite their rapid interspecies spread over the last few decades, their threat level has not gained a deserved high-profile reputation. − A subset of these enzymes mainly modify an antibiotic’s molecular target by catalyzing the transfer of a methyl group(s) to corresponding substrates using S-adenosylmethionine (SAM) as a cofactor. , Most members of this class modify the bacterial rRNA (rRNA) at strategic positions near the binding sites of various antibiotics, promoting resistance by indirectly inactivating protein synthesis inhibitors. The resulting modifications arguably cause steric hindrance, which is responsible for the observed resistance phenotype affecting several classes of antibiotics. − Erythromycin resistance methyltransferases (Erms) belong to this group of enzymes. They utilize a well-characterized mechanism to mono- or dimethylate adenosine 2058 (A ₂₀₅₈; E. coli numbering) of the 23S rRNA in bacterial ribosomes, facilitating resistance to macrolides, lincosamides, and streptogramin type-B antibiotics. , Remarkably, the strategic placement of the methylated nucleotides has no adverse impact on ribosome function, which might explain why this type of resistance is on the rise. Several Erm variants exist in clinical settings, magnifying the potential threat they pose, especially if they end up in bacterial strains with an already established multidrug resistance phenotype. ErmC is an example of a well-characterized methyltransferase, which was first discovered in 1971 and has since spread into multiple species of pathogenic bacteria, including S. aureus and Corynebacterium diphtheriae. , ErmC genes found in different species are highly conserved, suggesting that an inhibitor would likely have a broad-spectrum effect against this subclass of enzymes. , The limited number of studies investigating strategies for inhibiting this enzyme’s activity highlights the use of computational screening to examine small molecules or peptides as potential inhibitors. , Some of these studies further use minimum inhibitory concentration (MIC) data for comparative analysis to validate the biological application of such inhibitors. − These studies lack a more comprehensive biological characterization of the identified compounds, highlighting their impact on recovering the activity profiles of the affected antibiotic. For example, there is limited evidence of the ability of inhibitors to revive the bactericidal properties of the antibiotic in question. Such analyses would paint a more vivid picture of how inhibitors of this methyltransferase can be applied clinically to fight this form of resistance. The ever-expanding diversity of Erm variants recorded over the past few decades paints a dire picture of the bacterial evolutionary response to antibiotics, with no proper countermeasures in place to tackle this form of resistance. With so many identified variants from this subclass of enzymes, it is essential to devise ways of inhibiting their catalytic activity to rescue affected antibiotics and prevent an imminent health crisis.

Here, we present a comprehensive approach, combining computational and wet-lab methods, to characterize the impact of inhibiting ErmC on the antibacterial activity of affected antibiotics. We employ macromolecular solvent mapping and virtual screening to identify small-molecule binding hotspots and prospective inhibitors of ErmC. We use a constructed E. coli resistance working model expressing the S. aureus ErmC (SaErmC) variant to demonstrate substantial loss of activity for erythromycin (ERY) and clindamycin (CLN). Using this strain, we conducted single-dose combinatorial screens and identified test compounds with antibiotic-adjuvant profiles. We further establish that an identified noncompetitive inhibitor of SaErmC, JNAL-016, binds to the SAM binding pocket and exhibits promising adjuvant properties by terminating ErmC-mediated resistance, restoring the bactericidal properties of ERY. Our findings suggest that targeting resistance-conferring methyltransferases could be a viable strategy for developing adjuvant-based therapy against this resistance mechanism in pathogenic bacteria.

Results and Discussion

Computer-Aided Solvent Mapping and Virtual Screens Identify Potential ErmC Inhibitors

We initially employed a computational solvent mapping software (FTMap) to identify small-molecule binding hotspots using an available crystal structure of ErmC (PDB ID: 2ERC). The mapping algorithm uses standard structurally distinct probes to identify small-molecule binding pockets, located in consensus probe cluster sites. Using the FTMap server, we scanned the entire structure of this enzyme and identified a handful of predicted hotspots, identifiable by the clustering of probe molecules (Figures A and S1). Not surprisingly, one of the predicted binding hotspots with significant probe clustering represented the SAM binding site (Figure A), validating the accuracy of the mapping results. Some other key cluster sites were localized at locations associated with rRNA substrate binding, including the pocket where A ₂₀₅₈ usually docks for the methylation event (Figures A and S1). Encouraged by these results, we conducted computer-aided virtual screens to identify chemical scaffolds predicted to bind to the same crystal structure of ErmC. We were particularly interested in investigating ErmC over other variants because of the abundance of this allele in several species of the Staphylococcus and Bacillus genus, including its presence in clinical isolates of methicillin-resistant Staphylococcus aureus. − For this study, we focused on an antibacterial screening compound library from the Life Chemicals, Inc. repository, which contains small drug-like molecules designed to potentially have antibacterial properties.

1.

Solvent mapping and computer-aided screens predict compounds with high binding affinity to SaErmC. (A) Solvent mapping results performed using the FTMap algorithm, highlighting probe cluster regions (dotted sections) representing “hotspots” for potential small-molecule binding. The image shows the overlaid structure of an AlphaFold predicted model of the SaErmC sequence used in this study and the available cocrystal structure of SAM (yellow) bound to the B. subtilis ErmC (PDB ID: 1QAO). The green dotted hotspot represents the entry point for A ₂₀₅₈ to initiate the methylation event. Insetcofactor binding site with SAM highlighted in yellow surrounded by probe clusters, accurately depicting a binding pocket; (B) predicted binding energies for identified hits from a virtual screening experiment using AutoDock Vina with the search grid set to scan the entire protein structure. Compounds with a predicted energy ≤−10 kcal/mol were prioritized for secondary computational assessment.

We hypothesized that using such a library would produce inhibitors that could penetrate the bacterial membrane. As part of our strategy, we envisioned using any identified compound exhibiting antibacterial properties at a subgrowth-inhibitory dose for this study. The AutoDock Vina algorithm was used to conduct an unbiased screen of this library containing 10,880 compounds. The search grid encompassed the entire enzyme structure to increase the possibility of identifying potentially effective inhibitors based on their binding location. Results from these experiments identified compounds with an array of predicted binding energies, and we focused on those with ≤−10 kcal/mol (Figure B). This subset of compounds was re-evaluated using the Swiss ADME computational algorithm in a secondary screen for predicting desirable drug-like and pharmacokinetic features. We compared the compounds based on projected properties, including solubility, lipophilicity, pharmacokinetics, and druglikeness. This assessment enabled us to narrow the candidate list and select a subset of 20 compounds predicted to bind to the cofactor or substrate binding regions on the protein. We used this subset of compounds for the initial screen, seeking to develop a comprehensive strategy for inhibiting the activity of ErmC and related resistance methyltransferases using antibiotic–adjuvant combinations.

Expression of SaErmC in an E. coli Model Causes Resistance to Macrolides and Lincosamides

Variants of ErmC cause resistance against MLS_B antibiotics. To develop a working resistance model for screening potential inhibitors, we codon-optimized the gene sequence for SaErmC (UniProt) and cloned it into the tightly regulated arabinose-inducible pBAD24 expression vector. This plasmid was then transformed into an E. coli BW25113 strain with a tolC efflux pump deletion (BW25113tolC::cat), resulting in our resistance model strain, TSB-001 (Table S1). We confirmed the expression of SaErmC in this strain following induction with arabinose using SDS PAGE analysis (Figure S2). This strain was used to evaluate the resistance phenotype using representative antibiotics ERY and CLN from the macrolide and lincosamide classes, respectively. These antibiotics inhibit protein synthesis by binding to the 50S ribosomal subunit within the proximity of A ₂₀₅₈, hence the observed clinical resistance resulting from methylation of this nucleotide by ErmC variants. ,

Broth dilution assays with these two antibiotics in the presence or absence of SaErmC expression revealed dramatic changes in their activity profiles. For ERY, expression of this enzyme resulted in a 32- and 4.5-fold increase in the recorded MIC and IC₅₀ values, respectively (Figure A,B; Table ). CLN had a more dramatic change with a recorded 817- and 309-fold increase in its MIC and IC₅₀, respectively (Figure C,D; Table ), suggesting complete loss of activity under clinical settings. We conducted a control experiment with tetracycline (TET), an antibiotic that inhibits protein synthesis by targeting the accommodation site within the 30S ribosomal subunit. The binding site of TET is not within the proximity of A ₂₀₅₈, implying that its activity should be unaffected by the SaErmC methyltransferase activity. Indeed, the expression of SaErmC in strain TSB-001 revealed no notable changes in either the MIC or IC₅₀ values of TET (Figure E,F; Table ). These data suggest that the loss of activity observed for ERY and CLN in the two strains is directly linked to the activity of SaErmC, which catalyzes the methylation of A ₂₀₅₈ of rRNA within the 50S bacterial subunit. The data also confirmed the resistance phenotype of SaErmC-expressing strain TSB-001, enabling us to screen for potential adjuvants.

2.

SaErmC-mediated resistance is specific to MLS_B antibiotics. Results from growth inhibitory experiments conducted with or without SaErmC expression using broth microdilution assays to determine the minimum growth inhibitory concentrations (A,C,E) and corresponding dose response profiles (B,D,F), for ERY (A,B), CLN (C,D), and tetracycline (TET) (E,F) antibiotics. For the dose response plots, the data originate from cells cultured in the presence (filled squares) or absence (open circles) of SaErmC expression. The dotted arrows indicate a substantial loss in activity of the corresponding antibiotic observed for ERY and CLN but not TET. The data presented are averages from three biological replicates, with the error bars representing the standard deviations.

1. Bacterial Growth Inhibitory Activity Data for Antibiotics against Strain TSB-001 in the Presence or Absence of SaErmC Expression.

	MIC (μg/mL)			IC50 (μg/mL)
antibiotic	(−)SaErmC	(+)SaErmC	MIC fold change	(−)SaErmC	(+)SaErmC	IC50 fold change
CLN	1.3 ± 0	1062 ± 0	817	0.70 ± 1.7	216 ± 7.2	309
ERY	2.5 ± 0	80.00 ± 0	32	1.5 ± 2.0	6.80 ± 2.4	4.5
TET	0.37 ± 0.16	0.3700 ± 0.1600	1	0.16 ± 0.010	1.30 ± 0.01	1.3

^aThe activity data in the table are presented as the mean ± SD from ≥3 biological replicates.

The observed dramatic increase (∼817-fold) in the MIC of CLN when ErmC was expressed, relative to ERY (32-fold), is somewhat puzzling. Although more studies are necessary, we speculate that this observation may be due to the binding orientation of the two antibiotics within the bacterial ribosome and their proximity to the methylated N⁶ atom of A₂₀₅₈. Cryo-EM structures of CLN (PDB ID: 4V7V) and ERY (PDB ID: 7NSO) bound to the 70S ribosome reveal that the former binds significantly closer to A ₂₀₅₈, making direct polar contacts with N⁶, unlike ERY (Figure S3). Methylation of N⁶ by ErmC would terminate those interactions and introduce substantial steric hindrance, potentially interfering with the binding of CLN to the ribosome, as opposed to ERY (Figure S3).

Cell-Based Single-Dose Combinatorial Screens Identify Small-Molecule Candidates with Antibiotic–Adjuvant Properties

We initially set out to evaluate the potential antibacterial properties of the 20 compounds identified from our computational screen given that they originated from a library of molecules that might possess antibacterial properties. We conducted assays to determine any potential growth inhibitory activity using our SaErmC E. coli resistance model (Figure A). As presumed, all 20 compounds displayed varying levels of bacterial growth inhibition, some with recorded MIC values below 1 μg/mL (Table S2). Given that these small molecules were among those predicted to bind ErmC with high affinity, potentially inhibiting its catalytic activity, we opted to evaluate their potential adjuvant characteristics at much lower concentrations, displaying no considerable growth inhibitory activity for this study. Using our resistance model, we conducted a single-dose combinatorial screen with each compound at 0.02 μg/mL in the presence or absence of ERY (1.25 μg/mL) or CLN (50 μg/mL). Strain TSB-001 was highly resistant to the selected concentrations of both antibiotics, displaying 100% growth under induction conditions (Figure B,D). Under these conditions, some test compounds (JNAL-9, 11, 20, and 21) still exhibited growth inhibition when tested alone (Figure B). Regardless, the overall trend for most of the compounds revealed a strong synergistic effect when combined with ERY (Figure C) or CLN (Figure D). The combinatorial activity profiles of some test compounds, such as JNAL-10, 12, and 16, were particularly encouraging. As a result, we selected six compounds (JNAL-3, 9, 10, 16, 20, and 21; Table S3) with varying activity profiles and determined an ideal dose for combinatorial assays. When the growth inhibitory profile of ERY was examined in the presence or absence of fixed doses of these prospective adjuvants using our resistance model, all six ERY–inhibitor combinations reduced the MIC and IC₅₀ values of the antibiotic by 2–5.7-fold, relative to ERY-only values (Table ). The same concentrations of adjuvants decreased the MIC and IC₅₀ values of the CLN by 1.4–5.7-fold (Table S4). These results demonstrate promising antibiotic-adjuvant properties for the test compounds against SaErmC-mediated resistance. JNAL-016 was particularly attractive given its profile in the initial single-dose screen (Figure B–D) and its synergistic effect on the MIC of ERY and CLN (Tables and S4, respectively), warranting further characterization.

3.

Cell-based single-dose screen identifies compounds with potential adjuvant properties. (A) Schematic showing the design of the cell-based assay used to mimic SaErmC-mediated resistance for screening prospective antibiotic–adjuvants. Promising adjuvants were selected based on their ability to act synergistically with the corresponding antibiotic and inhibit bacterial growth when expressing SaErmC. (B–E) Heat maps showing results from the single-dose screening experiments for 20 compounds using the E. coli resistance model under SaErmC expression conditions. The test compounds were each evaluated at a concentration = 0.02 μg/mL. The scales represent the percentage relative growth with the negative control (N-con; red) and the positive control (P-con; cyan) depicting maximum growth and complete inhibition, respectively. The heat-map data highlights the growth inhibitory properties of the test compounds evaluated alone (B) or in combination with ERY (1.25 μg/mL) (C) or CLN (50 μg/mL) (D) at concentrations where the strain exhibits complete resistance. (E) The identities and location of the test compounds and assay controls in each test plate used to generate the heat map. Each grid square represents one sample, and the data shown are averages from three-well replicates. Wells G3 and H3 have no samples. See Table S1 for pTSB-001 description.

2. Antibacterial MIC and IC₅₀ Profiles of ERY in the Presence or Absence of Candidate Adjuvants against Strain TSB-001 Expressing SaErmC .

adjuvant information		ERY MIC (μg/mL)			ERY IC50 (μg/mL)
ID	concentration used (μg/mL)	(−) adjuvant	(+) adjuvant	MIC fold change	(−) adjuvant	(+) adjuvant	IC50 fold change
JNAL-003	0.058	80 ± 0	40 ± 0	2.0	6.8 ± 2.4	3.1 ± 1.4	2.2
JNAL-009	0.025		20 ± 0	4.0		1.6 ± 1.2	4.3
JNAL-010	0.010		20 ± 0	4.0		2.9 ± 0.7	2.3
JNAL-016	0.042		20 ± 0	4.0		1.2 ± 0.4	5.7
JNAL-020	0.022		15 ± 6	5.3		2.1 ± 0.7	3.2
JNAL-021	0.022		18 ± 5	4.6		1.7 ± 0.9	4.0

^aThe activity data in the table are presented as the mean ± SD from 3 biological replicates.

Prospective Adjuvants Display No Detectable Cytotoxicity against Human Embryonic Kidney Cells

To assess potential toxicity against mammalian cells, we examined the cidal activity of the six promising adjuvant candidates against human embryonic kidney cells (HEK293) using a propidium iodide (PI) fluorescence permeability cell viability assay. PI is a charged fluorescent dye impermeable to healthy bacterial membranes but can enter membrane-compromised cells and bind to nucleic acids, enhancing its fluorescence, facilitating cell-death quantification. , In this assay, the test compounds were examined by treating cells with concentrations of 20 times their corresponding MIC values (Table S2). The data obtained was quantified relative to negative control samples, wells containing a cell lysis reagent, which was used to depict 100% death in the assay. The results showed no toxicity for any of the six compounds at the dose examined against HEK293 cells (Figure S4). The basal fluorescence detected for all the compound-treated samples was comparable to that of the DMSO-treated samples (≤5%). These results provide a promising outlook for developing such compounds for use as adjuvants against SaErmC-mediated resistance to MLSB antibiotics because they would provide a large therapeutic window for administering antibiotic–adjuvant combination doses.

JNAL-016 Abolishes the Resistance Mediated by SaErmC against ERY

For all follow-up experiments, we used JNAL-016 in this proof-of-concept study. This compound was among those predicted to have the highest binding affinity for SaErmC from computational data and displayed desirable antibiotic–adjuvant properties in our preliminary cell-based experiments highlighted above (Figure B–D; Tables and S4). We considered the possibility that JNAL-016 might interfere at the transcription or translation level to prevent the expression of the enzyme, indirectly terminating resistance. We quickly disproved this probability, given that the enzyme was expressed at a comparable level in the presence or absence of this test compound (Figure S2). To demonstrate the impact of JNAL-016 on the activity of ERY under SaErmC-mediated resistance conditions, we conducted a set of growth inhibitory experiments with and without induction of the enzyme in the presence or absence of this compound. The addition of JNAL-016 at 0.042 μg/mL to broth dilution growth inhibitory assays was enough to completely resensitize strain TSB-001 expressing SaErmC to ERY (Figure A), terminating the resistance phenotype. However, although the same dose of the adjuvant displayed promising activity when combined with CLN, it did not fully restore sensitivity to this antibiotic (Figure S5). This result suggests the need for a higher dose of the adjuvant to overcome the >300-fold increase in the recorded IC₅₀ of CLN when the enzyme is expressed (Figure ; Table ). Regardless, these data are promising and illustrate the potential use of adjuvants to tackle SaErmC-mediated resistance. To further characterize the activity profile of JNAL-016, we complemented this study by conducting growth inhibitory experiments using antibiotic-infused test strips (Liofilchem) on a solid agar medium. LB-agar plates were prepared by adding strain TSB-001 in the presence or absence of arabinose to the molten top agar. JNAL-016 (0.25 μg/mL) was added to a subset of plates with SaErmC expression. We then placed ERY-, CLN-, or TET-infused test strips onto designated plates to evaluate the potential adjuvant properties of JNA-016 under these growth conditions. Our results confirmed resistance of this strain toward ERY and CLN when SaErmC was expressed, depicted by a high level of bacterial growth around the test strips at higher antibiotic levels, resulting in elevated MIC values (Figure B,C; Figure S6; Table S5). The addition of JNAL-016 in the presence of the enzyme reverted the activity of ERY and CLN to mimic those of the antibiotic-sensitive cultures, depicting similar trends to our broth dilution assays (Figure B,C; Figure S6; Table S5). Importantly, our control experiment demonstrated that the activity of TET was not affected by the presence of SaErmC or the activity of JNAL-016 (Figure B,C; Figure S6; Table S5), a result that also confirms the specificity of the Erm-mediated resistance to MLS_B antibiotics. We also conducted checkerboard assays to more accurately quantify the synergistic relationship of ERY and CLN when combined with JNAL-016. The cell density data obtained were analyzed using an isobologram and revealed a strong synergistic relationship, whereby 0.03 μg/mL of the adjuvant enhanced the MIC of ERY 16-fold or 4-fold for CLN (Figure D). This synergistic effect was also evident in the heat maps generated from the cell density data of the checkerboard assays (Figure S7A,B). Interestingly, the frequency of spontaneous mutants resistant to a combination of ERY (50 μg/mL) and JNAL-016 (0.042 μg/mL) was <5.1 × 10^–8, compared to 5.9 × 10^–1 for ERY alone (50 μg/mL), suggesting that the emergence of resistance to an ERY-JNAL-016 administration would be rare.

4.

JNAL-016 displays promising adjuvant properties against SaErmC-mediated resistance when combined with ERY or CLN. (A) Growth inhibitory dose response profiles for ERY assessed in the absence (green triangles) or presence (red squares) of SaErmC expression and in the presence of both SaErmC and JNAL-016 (0.042 μg/mL) (blue circles). (B) Determination of the MICs using antibiotic-infused strips. For induction and samples treated with the adjuvant, arabinose (1%) (middle column) or a combination of arabinose and JNAL-016 (last column) was added to the molten top agar before solidifying it on the plate and placing the strips. The compound concentration did not inhibit growth independently, as evident in the observed bacterial growth at the bottom of the strips. (C) Graph showing the quantified MIC data from the top-agar experiments. The MICs were determined in the absence (green bars) or presence (red bars) of SaErmC induction and with an induced set of samples containing JNAL-016 at 0.25 μg/mL (blue bars). For graphs A and C, the data presented are averages from three biological replicates, with the error bars representing the standard deviations. (D) An isobologram profiling the fractional MIC (FMIC) results from checkerboard assays, highlighting the strong synergistic relationship between JNAL-016 and ERY (rectangle) or CLN (triangle) in combination. Only 0.03 μg/mL JNAL-016 was needed to reduce the MIC for ERY by 16-fold (4-fold for CLN). Average data from three biological replicates are presented.

JNAL-016 Displays Adjuvant Properties against WT BW25113 Expressing SaErmC

We conducted most of our experiments in this study using efflux-deficient strain TSB-001 (Table S1). However, considering that clinical isolates of E. coli carrying the ErmC gene would have an active TolC efflux pump, we wanted to establish whether JNAL-016 would retain its adjuvant properties in a wild-type (WT) strain of BW25113, TSB-014, expressing the enzyme (Table S1). Initially, we determined the MIC and IC₅₀ values for the antibiotics in this strain under induction or repression conditions, and as expected, the values increased dramatically under both settings (Table S6). For example, under ErmC expression, the MIC for ERY against strain TSB-014 increased 100-fold and that for CLN was 2.5-fold relative to strain TSB-001 (Tables and S6). Checkerboard assays with JNAL-016 revealed a synergistic profile for both antibiotics similar to that observed in the efflux-deficient strain, albeit requiring higher concentrations of the adjuvant (Figure S7C,D). For instance, 10 times the concentration of JNAL-016 used to abolish resistance against ERY in the efflux-deficient strain was required to decrease the MIC of this antibiotic 4-fold or 2-fold for CLN (Tables and S6). These findings are promising when considering clinical isolates of deadly pathogens exhibiting ErmC-mediated resistance. However, follow-up investigations are warranted to fully characterize the activity of this promising adjuvant against clinical isolates.

Time-Kill Assays Confirm the Adjuvant Properties of JNAL-016 and Its Ability to Restore the Bactericidal Activity of ERY

To test the activity of antibiotic–JNAL-016 combinations in actively growing cultures, we employed a PI-based fluorescence detection system in our time-kill assays. For these experiments, PI (5 μM) was added to the culturing medium and we monitored the synergistic effects of JNAL-016 on the activity of ERY using cultures expressing SaErmC to mimic the resistance phenotype. Different antibiotic concentrations were evaluated when combined with a fixed dose of JNAL-016 (0.085 μg/mL) in cultures grown at 37 °C in a dry air incubator shaking at 325 rpm for 8 h. During this period, we measured bacterial growth by recording the cell density (OD_600 nm) and estimated viable cells from the emitted PI fluorescence every hour. To calculate cell death under each concentration of ERY used, we normalized the obtained PI fluorescence by dividing those values by the total cell density to eliminate unreliable data. This strategy accounted for any observed variability in the fluorescence signal resulting from differential growth rates in samples treated with lower vs higher ERY concentrations. When comparing bacterial growth in the presence or absence of JNAL-016, our experiments demonstrated substantial growth inhibition in samples treated with ERY–JNAL-016 combinations relative to those containing ERY alone (Figure A,B and S8). This phenomenon was particularly more apparent in samples containing ≤20 μg/mL ERY, where ≥40% growth (relative to the DMSO control) was recorded in the ERY-only treated samples (Figure A). In contrast, the presence of JNAL-016 in samples containing the same range of ERY concentrations showed complete growth inhibition over the 8 h period, except for the later stages of samples containing 0.625 or 1.25 μg/mL of ERY, which had elicited some growth (Figure B). Cell-death profiles had a similar trend, revealing several ERY–JNAL-016 combinations with enhanced cell killing, depicted by an increase in PI fluorescence relative to the DMSO controls (Figure C,D; Figure S8). The absence of JNAL-016 resulted in varying levels of cell death, proportional to the concentrations of ERY tested (Figure C). Like the growth profiles, ERY concentrations of ≤20 μg/mL caused the least cell death. However, drastic changes were observed in JNAL-016-treated cultures. While antibiotic–adjuvant combinations involving ERY concentrations ≥40 μg/mL had a similar trend to the ERY-only samples, lower doses of the antibiotic demonstrated the desired adjuvant properties of JNAL-016, enabling ERY to overcome SaErmC-mediated resistance (Figure D). Based on the PI cell death quantification, the combined antibiotic–adjuvant effect led to substantial cell death, with ∼90% (1 log unit increase in the fluorescence) reduction in the viable cell count recorded at 2.5 μg/mL ERY after 8 h (Figure D).

5.

JNAL-016 acts as an adjuvant to restore the bactericidal properties of ERY in time-kill assays. (A,B) Growth profiles of strain TSB-001 expressing SaErmC treated with varying concentrations of ERY monitored by cell density (OD_600 nm) every hour over 8 h in the absence (A) or presence (B) of JNAL-016 (0.085 μg/mL) added to every sample. (C,D) Time-kill profiles for the same samples in “A and B” presented as cell-density-normalized PI fluorescence over 8 h in the absence (C) or presence (D) of JNAL-016. Samples with higher fluorescence on the graphs have more dead cells. (E,F) Quantification of the colony-forming units from the same samples in “A and B” performed at different time points and concentrations of ERY in the absence (E) or presence (F) of JNAL-016. DMSO control cultures for samples in panels (A,C,E) indicate no ERY present in the sample, while those in panels (B,D,F) are treated with only JNAL-016 (dissolved in DMSO) at 0.085 μg/mL. For graphs A–D, the data presented are averages from three technical replicates, and the error bars represent the standard deviation of the mean. Additional biological replicates for the time-kill assays are presented in Figure S8.

To further highlight the restored bactericidal properties of ERY when combined with JNAL-016, we complemented this fluorescence cell viability determination with plating experiments on solid agar to determine the viable cell counts. Sample aliquots from ERY or ERY-JNAL-016 combination treatments in the same time-kill assays highlighted above were extracted at 0, 4, and 8 h time points, diluted accordingly, and plated on LB agar. Viable cell counts were quantified after a 24 h incubation period at 37 °C. The observed trends from the analyzed samples mimicked those observed in the growth inhibition and PI-linked death profiles (Figure E,F; Figure S8). When ERY was used alone, 80 μg/mL was the only dose sufficient to kill >99% of the viable cells after 8 h. However, this concentration is 32-fold higher than the recorded MIC for this antibiotic in the absence of the enzyme, making it unrealistically high for typical experiments (Figure E and Table ). At physiologically relevant concentrations, the number of viable cells increased gradually with time, and there was no notable change in the cell counts by the 8 h time point for cells treated with ≤10 μg/mL of the antibiotic (Figure E). In contrast, the addition of JNAL-016 (0.085 μg/mL; a 2× MIC dose was required in active cultures) to the antibiotic-treated cultures led to considerable cell death after 8 h for ERY concentrations of ≤10 μg/mL (Figure F). For example, 2.5 μg/mL of ERY in combination with the adjuvant reduced the number of viable cells by >99% (Figure F). This concentration is significant because it mimics the recorded MIC for ERY in a strain lacking SaErmC expression in this study (Table ), confirming the restored bactericidal properties of this antibiotic. Taken together, results from the time-kill assays establish JNAL-016 as a promising candidate for adjuvant-based therapy against SaErmC-mediated resistance.

JNAL-016 Is a Noncompetitive Inhibitor of SaErmC Activity In Vitro

While JNAL-016 appears to have an alternative antibacterial mode of action at high concentrations, our biological data suggest that when combined with ERY or CLN, it acts as an adjuvant against SaErmC-mediated resistance at lower doses, lacking independent growth inhibitory activity. To characterize the biochemical interaction and potential inhibitory profile of JNAL-016 against this enzyme, we developed an expression strain for purifying recombinant SaErmC. The gene encoding SaErmC was cloned into the pET28a expression vector, which was transformed into BL21­(DE3)­pLysS cells, resulting in strain TSB-010 (Table S1). Expression and purification of SaErmC from this strain yielded a high-purity protein (Figure A), which was used to characterize the inhibition kinetics of JNAL-016 using the MTase-Glo methyltransferase assay (Promega). A 32-oligonucleotide (32 nt) RNA fragment from 23S rRNA, previously shown to be a substrate for different Erm variants, , was used to develop an inhibitor screening assay. For this experiment, methylation of the 32 nt RNA substrate catalyzed by SaErmC using SAM would generate S-adenosylhomocysteine (SAH), which is required as a substrate for the reaction that ultimately generates detectable luminescent light. Initially, we confirmed the linear correlation between the concentration of SAH and the emitted light, which facilitated the quantification of our experimental data (Figure S9). The activity of SaErmC and closely related methyltransferases depends on the universal methyl donor SAM for catalysis. As a result, it is feasible to project that this molecule could have copurified with our enzyme during purification, potentially interfering with our data analysis. We therefore assessed the purified SaErmC for possible contamination with copurified SAM by conducting MTase-Glo activity assays under various conditions, including the absence or presence of exogenously added cofactor (Figure B). Results from this experiment showed very little catalytic activity (∼5%) in the absence of exogenously added SAM compared to that when this cofactor was added to the reaction (Figure B). We then assessed the steady state kinetics of SaErmC catalysis and the corresponding inhibition by JNAL-016 using RNA substrate concentrations ranging from 0 to 8 μM over a 30 min incubation period. SaErmC catalyzed the methylation of the 32 nt RNA substrate with a V _max = 7.6 (±0.42) × 10^–10 M s^–1 and K _m = 3.4 (±0.11) × 10^–6 M (Figure C and Table ). The addition of JNAL-016 resulted in a 2.1-fold reduction in the reaction rate (_app V _max = 3.7 (±0.81) × 10^–10) but did not alter substrate binding (_app K _m = 3.4 (±0.97) × 10^–6 M) (Figure C and Table ). Accordingly, this compound decreased the turnover and catalytic efficiency of SaErmC by ∼2-fold (Table ). A comparison of the kinetic data in the presence or absence of this inhibitor suggested a noncompetitive mode of inhibition. These data demonstrate that JNAL-016 is an effective inhibitor of SaErmC activity in vitro, supporting the biological data highlighting its adjuvant properties when combined with ERY and CLN against SaErmC-mediated resistance.

6.

JNAL-016 is a noncompetitive inhibitor of SaErmC. (A) SDS polyacrylamide gel showing samples from the purification of recombinant SaErmC following expression in BL21­(DE3) cells. L1protein marker; L2uninduced sample; L3induced sample; L4clarified lysate; L5flowthrough from Ni-NTA binding; L6wash fraction after affinity binding; L7pooled elution fractions after affinity purification; L8purified SaErmC after gel filtration on an S200 column (Cytiva). (B) Characterization of the purified enzyme using the MTase-Glo assay to determine the presence of copurified SAM and potential nonenzymatic autohydrolysis of this cofactor under our experimental conditions. The data show no autohydrolysis of SAM in the absence of the RNA substrate, and copurified SAM accounts for ∼5% of the recovered luminescence signal. (C) Michaelis–Menten plots showing the steady state kinetics of SaErmC catalysis for the methylation of a 32-mer RNA substrate in the absence (open circles) or presence (filled squares) of JNAL-016. The data points represent averages from three independent experiments ± SD. These data suggest a noncompetitive mode of inhibition by the inhibitor.

3. Michaelis–Menten Steady State Kinetic Data for SaErmC in the Presence or Absence of JNAL-016.

sample ID	K m (M)	V max (M s–1)	K cat (s–1)	K cat/K m (M–1 s–1)
(−) JNAL-016	3.4 (0.11) × 10–6	7.6 (0.42) × 10–10	1.5 × 10–3	4.5 × 102
(+) JNAL-016	3.4 (0.97) × 10–6	3.7 (0.81) × 10–10	7.3 × 10–4	2.2 × 102

^aThe data presented are the mean values from three independent experiments (SD).

JNAL-016 Competes with SAM, But Not the Substrate, for the Same Binding Pocket on SaErmC

The steady state kinetic data for SaErmC in the presence of JNAL-016 suggested that the compound could bind at an allosteric site to inhibit catalysis. This observation prompted structural studies attempting to cocrystallize an enzyme–compound complex to gain further insight into the binding location and mechanism of inhibition. Although the crystallization attempts have been unsuccessful, we took a different approach that utilized binding experiments to better understand the potential enzyme–inhibitor interactions. We used microscale thermophoresis to conduct a binding experiment titrating JNAL-016 into a fixed enzyme concentration and obtained a binding profile for this compound, which suggested a tight interaction (K _d = 1.8 ± 0.7 μM) (Figure A). To gain further insight into where this inhibitor might be binding, we considered its established noncompetitive mode of inhibition of the SaErmC activity. We then hypothesized that JNAL-016 could either compete with the SAM for its binding pocket or bind at an alternative allosteric site that indirectly blocks cofactor access to the enzyme. To test this hypothesis, we conducted binding competition experiments using microscale thermophoresis, where a fixed concentration (2 μM) of 50-[N-[(3S)-3-amino-carboxypropyl]-N-methylamino]-50-deoxyadenosine (aza-SAM) was added to serially diluted samples of JNAL-016 prior to adding the enzyme. The labile methyl-sulfonium ion of SAM made it unstable under our binding experimental conditions. As such, we used a more stable analog, aza-SAM, which contains a nonextractable aminomethyl replacement. This analog has been used in the past to replace SAM in crystallography and other reactions involving methyltransferases. Under these conditions, the enzyme–JNAL-016 interaction decreased substantially, resulting in an ∼57-fold decrease in the binding affinity (_app K _d 122 ± 13 μM) (Figure B). This result suggested a competitive event for the SAM binding site, which reduced the available binding sites for JNAL-016.

7.

JNAL-016 competes with SAM for the cofactor binding site. Results from in vitro experiments highlighting the binding interaction between aza-SAM and JNAL-016 were determined using microscale thermophoresis. (A) The binding profile of JNAL-016 (K _d = 1.8 ± 0.7 μM) to purified SaErmC (100 nM). (B) Results from binding experiments to SaErmC showing the profile of JNAL-016 alone (open circles) and in the presence of 2 μM aza-SAM (_app K _d = 122 ± 13 μM; filled squares). (C) Binding profiles for JNAL-016 alone (open circles) and in the presence of 45 μM 32 nt RNA substrate (_app K _d = 0.9 ± 0.2 μM; filled squares). Each data set represents averages of three replicates, and the error bars represent the standard deviation of the mean. (D) Cartoon-surface representation of ErmC showing the docking results of JNAL-016 overlaid with a structure containing bound SAM within the cofactor binding pocket. The structure shows two different predicted conformations of the inhibitor, highlighted in cyan and magenta. (E) Close-up view of the cofactor binding site showing the distinct predicted binding orientation for JNAL-016 relative to SAM. The top image shows a transparent surface view revealing the predicted conformations of the enzyme’s structure. The bottom image depicts the location of the binding pocket with a section of the molecule blocking the entry point for nucleotide A ₂₀₅₈ during the methylation event, suggesting a possible inhibition strategy for the compound. (F) Close-up showing a cartoon structure of ErmC docked with JNAL-016, highlighting the predicted polar contacts (yellow dashed lines) with specific residues of the enzyme, including N101, which is essential for substrate methylation during catalysis.

To confirm that JNAL-016 did not inhibit SaErmC catalysis competitively, we conducted a similar experiment, adding 45 μM 32 nt RNA substrate instead of aza-SAM. Unlike the effect observed with the cofactor, the presence of the RNA appeared to stabilize JNAL-016 binding, resulting in a higher affinity (_app K _d = 0.9 ± 0.2 μM) (Figure C). This observation implied that this compound was not competing with substrate binding, further supporting our enzyme inhibition kinetic data (Figure C). It is noteworthy that the latter observation could result from the large binding surface area of the substrate and the multiple enzyme–RNA contact points, , which may hinder a competitive binding event with JNAL-016. Regardless, the presence of the RNA substrate resulted in a ∼2-fold enhancement in the affinity of JNAL-016 for SaErmC. In silico molecular docking experiments of this inhibitor to an AlphaFold-modeled SaErmC structure localized it to the SAM binding site, with a section of the molecule potentially blocking the entry point of the adenine (A ₂₀₅₈) to be methylated by the enzyme (Figure D–F). While a blockade at this location might result in competitive inhibition for a typical enzyme substrate, we propose that the large enzyme–substrate contact surface area would allow the substrate to remain tightly bound without being able to dock the nucleotide into the active site (Figures S1 and S10). Also, binding to the enzyme in the predicted conformation (Figure E) might explain the observed competition with SAM for the cofactor binding region. The predicted binding location for JNAL-016 has several implications, including its potential ability to act as a broad-spectrum adjuvant against ErmC variants. This proposal is based on the observation that currently identified ErmC variants have a significantly high sequence and structural conservation, especially when considering residues that make polar contacts with SAM within the cofactor binding pocket, including those that are essential for enzyme function, like N101, or ones that facilitate substrate binding (Figure F). Taken together, these results further support our kinetic inhibition data, demonstrating that JNAL-016 inhibits the activity of SaErmC noncompetitively. These findings also make a strong case for developing and using adjuvants as broad-spectrum inhibitors for tackling enzyme-mediated antibiotic resistance by ErmC variants.

Conclusion

In this study, we investigated the use of small molecules to neutralize the activity of a variant of ErmC, which mediates resistance to MLS_B antibiotics in several pathogenic strains of bacteria. We successfully employed computer-aided solvent mapping and virtual screens to identify small-molecule binding sites and potential inhibitors of SaErmC. Using a developed E. coli SaErmC resistance model, we identified compounds with promising antibiotic–adjuvant properties against this form of resistance. Among these inhibitors, JNAL-016 displayed exceptional adjuvant activity when combined with ERY, enabling this antibiotic to regain its bactericidal properties, killing >99.9% of viable bacteria after 8 h. Our data characterizing this compound suggest that it binds tightly to SaErmC, competing with SAM for the cofactor binding site to inhibit this enzyme noncompetitively. This comprehensive study offers a compelling case for prioritizing similar inhibitors to target resistance-conferring methyltransferases using antibiotic–adjuvant-based combination therapy to tackle deadly bacterial pathogens such as S. aureus. The rapid spread of resistance-conferring methyltransferases like Erms in pathogenic strains of bacteria poses a significant threat to our ability to treat such infections. Given that the activity of such enzymes confers devastating resistance to clinically active antibiotic classes, it is essential to devise overarching countermeasures to deal with such cases. While this study may offer hope of a resolution, several key questions remain unanswered, warranting further investigation, including the potential clinical applications of Erm inhibitors or whether inhibitors such as JNAL-016 could have broad-spectrum activity against other variants of this class of methyltransferases. Ongoing research will investigate how such compounds, and additional scaffolds, interact with these methyltransferases, providing vital insight into the design of more effective inhibitors with superior adjuvant properties.

Methods

Computational Solvent Mapping

Solvent mapping and identifying small-molecule binding sites were conducted remotely using an available structure of ErmC (PDB ID: 2ERC) on the FTMap servers (Boston University; http://ftmap.bu.edu/) as previously described.

In Silico Virtual Screening and Molecular Docking

Computational virtual screens were conducted using the AutoDock Vina computational software. The virtual antibacterial screening library containing 10,880 compounds was obtained from the Life Chemicals, Inc. repository, which was retrieved from the database as a compressed structural data file. Open Babel was used for energy minimization of the ligands and file conversion to the required pdbqt format. Docking screens were performed with the search grid covering the entire ErmC structure (PDB ID: 2ERC), and the binding energies were used to select initial candidates. Prioritized compounds were further analyzed using the Swiss ADME (absorption, diffusion, metabolism, and excretion) computational software for suitable drug-like properties. Molecular docking with JNAL-016 onto ErmC (PDB ID: 2ERC) was conducted as previously described. The optimal conformation(s) with the lowest binding energy and rmsd value = 0 were selected.

Construction of the E. coli SaErmC Resistance Model and Expression Strains

The sequence of the S. aureus ermC gene (UniProt ID: P02979) in a pUC18 (or pET28a) vector was obtained from Gene Universal Inc., PCR amplified, and subcloned into the pBAD24 vector at 5′-EcoRI/3′-Xbal sites (FWD primer: 5′-ATATATGAATTCATGAATGAAAAGAACATTAAACAC; REV primer: 5′-ATATATTCTAGACTATTTATTGAACAGTTTGTACGAAT). The amplified segment was digested and ligated into the expression vector, followed by transformation first into DH5α cells, and finally into BW25113ΔtolC::cat cells, resulting in strain TSB-001. The SaErmC purification strain, TSB-010, was constructed by transforming the IPTG-inducible pET28a-SaermC into BL21 (DE3) cells.

Determination of the Minimum Inhibitory Concentrations (MICs) and Half-Maximal Growth Inhibitory Concentrations (IC₅₀s)

All of the broth screening assays followed the same procedures. Bulk working cultures were prepared by diluting overnight cultures of strains TSB-001 or TSB-014 in lysogeny broth (LB) containing 100 μg/mL AMP (final OD_600 nm 0.002). 2-fold serial dilutions of antibiotics or test compounds were prepared in the same medium containing 1% arabinose, if expression of SaErmC was required. Broth microdilution assays were performed as previously described. MIC values were determined by a visual inspection of the plate to identify the lowest concentration without visible growth. The half-maximal inhibitory concentration (IC₅₀) values were calculated from the dose–response plots of the OD_600 nm readings from a SpectraMax iD5 plate reader (Molecular Devices). These data were normalized to “drug-free” controls to generate dose–response plots. Triplicate data sets from ≥3 biological replicates were conducted and analyzed using GraphPad Prism.

Single-Dose Screening of ErmC Inhibitory Test Compounds

Dose–response growth-inhibition experiments were initially performed to determine the optimal concentration of the test compounds for use in the single-dose combination experiment with the antibiotics. The obtained dose–response plots were used to determine the concentration to use in the experiment, which corresponded to the concentration at which most of the test compounds did not exhibit significant antibacterial activity. A working stock culture of strain TSB-001 in LB with 2× the desired final cell density was prepared as described above and supplemented with 1% arabinose. Test compounds (final 0.02 μg/mL) were added to the same triplicate wells in three plates: one with no antibiotic, one containing 1.25 μg/mL ERY, and one with 50 μg/mL CLN, and the plates were incubated at 37 °C for 18 h, and the OD_600 nm was recorded as described. The data were normalized and quantified in GraphPad Prism to generate heat maps.

MIC Determination Using Antibiotic Strip Diffusion Assays

A 100 μL aliquot from a saturated overnight culture was added to molten top agar at 44 °C containing 100 μg/mL AMP, 1% arabinose or glucose (repression), and JNA-016 (0.25 μg/mL), when necessary. The top agar was poured onto prewarmed LB-AMP plates and left to solidify at room temperature before addition of the antibiotic test strips (Liofilchem). Plates were incubated at 37 °C for 20–24 h, and MICs were determined based on growth clearing according to the manufacturer’s directions. Data from three biological replicates were analyzed.

Glossary

Abbreviations

SAM

S-adenosylmethionine

SAH

S-adenosylhomocysteine

Aza-SAM

50-[N-[(3S)-3-amino-carboxypropyl]-N-methylamino]-50-deoxyadenosine

Erm

erythromycin resistance methyltransferase

ERY

erythromycin

CLN

clindamycin

TET

tetracycline

PI

propidium iodide

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acsinfecdis.5c00865.

• Supplemental experimental procedures; tables and figures highlighting additional results for computational experiments, cytotoxicity screens, antibacterial combinatorial assays, and protein expression; and materials used and compound identities (PDF)

J.N.A. conceived the study. J.N.A. performed cytotoxicity screens. T.S.B. and J.N.A. designed the experiments, analyzed the data, and prepared and wrote the manuscript.

The authors declare no competing financial interest.