The three-dimensional (3D) structures of the target proteins, including AdeA and AdeB, were retrieved from the RCSB Protein Data Bank (https://www.rcsb.org/) and UniProt (https://www.uniprot.org/). Protein structures were prepared by assigning appropriate bond orders and formal charges and by correcting missing hydrogen atoms and/or bonds to ensure reliability in docking simulations. All water molecules and non-essential heteroatoms were removed, except those critical for ligand–protein interactions. Ligand structures were obtained from chemical databases such as PubChem (https://pubchem.ncbi.nlm.nih.gov/) and downloaded in SDF format. Ionization states and relevant tautomeric forms were generated at physiological pH (7.4). The Epik tool was used to rank ligand conformations based on predicted pKa values, enabling identification of low-energy states that significantly influence docking accuracy. The prepared ligands were finally saved in “.mae” format for Glide docking (Gopikrishnan et al., 2023).

Protein binding pockets were identified using the CASTp (http://sts.bioe.uic.edu/castp/) and CASTpFold (https://cfold.bme.uic.edu/castpfold/) servers (Tian et al., 2018). The most suitable binding sites were selected based on the presence of druggable residues, as well as pocket surface area and volume. A docking grid box was generated using the grid generation tool, with interacting residues from the selected binding pocket manually specified to accurately define the grid. Molecular docking was subsequently carried out using the Schrödinger Protein Docking Tool (Gopikrishnan et al., 2023; Verliani et al., 2024).

Molecular docking was performed using Glide (Schrödinger Suite), employing the extra precision (XP) method for the selection of optimal confirmations (Bhati et al., 2025), and the results were validated by Glide scores, a numerical estimate of binding affinity. Docking algorithms and settings included multiple runs, sampling various binding poses, and reporting Glide scores and docking scores (kcal/mol) for each ligand. For comparative analysis, docking runs were repeated for the all-tested polyphenols.

The top-scoring complexes were visualized in Maestro and PyMOL (www.pymol.org), emphasizing hydrogen bonds, hydrophobic contacts, salt bridges, and π–π and anion–π interactions. Interacting residues, such as VAL14, ASN15, ILE16, GLN75, VAL78, GLN79, ILE82, LYS83, GLU86, LEU98, and LEU99, were documented and compared across ligands. A table summarizing Glide score, docking score, and interacting amino acids for each complex provided a way of choosing a suitable lead compound for further investigation. The 2D interactions were obtained from Maestro Schrödinger Software Suites, Maestro version 12.9 (Pauly et al., 2022), and the 3D interactions between ligands and proteins were visualized using PyMOL, version 3.1.6.1 (Yuan et al., 2016).

Molecular dynamics (MD) simulations were conducted using the GROMACS program with an AMBER-99SB force field (Wang et al., 2004). The simulations utilized the AdeB protein structure, with the ligand’s structure and electrostatic properties generated by ACPYPE. A dodecahedron box was fabricated, with the structural complex positioned at its center. The box was subsequently filled with TIP3P water molecules using the procedure described by Jorgensen et al. (1983). A minimum distance of 1 nm was established between the protein molecule and the boundary of the simulation box. GROMACS software application added counter-ions to achieve charge neutrality. A distance threshold of 14 Å was employed for non-covalent Van der Waals interactions. The LINCS method was employed to constrain covalent bonds involving hydrogen atoms (LINCS: A Linear Constraint Solver for Molecular Simulations - Hess - 1997 - Journal of Computational Chemistry - Wiley Online Library, 1998).

Energy was minimized using a step size of 0.001 nanoseconds. Afterward, a 100-picosecond simulation was performed in the isothermal–isovolumetric ensemble (NVT). Subsequently, a 10-nanosecond simulation was conducted using the isothermal–isobaric ensemble (NPT) to equilibrate the water system (Abraham et al., 2015). A molecular dynamics simulation was conducted using a non-polarizable tight-binding model. The simulation involved a 100-nanosecond production run. The simulation used a time step of 2 femtoseconds. The simulation employed a Parrinello–Rahman barostat and a modified Berendsen thermostat, maintaining a constant temperature of 300 K and pressure of 1 atm (Parrinello and Rahman, 1981). The root mean square deviation (RMSD) of the trajectory was calculated using GROMACS 2023.1 tools.

MIC reversal assays for antibiotics in the presence and absence of polyphenols and standard efflux pump inhibitor PAβN (RND pump inhibitor) were performed for all A. baumannii clinical isolates to screen for the effective compound, which reversed the MIC of antibiotics in drug-resistant clinical isolates. In short, polyphenols/PAβN at sub-MIC concentrations were incubated with varying concentrations of erythromycin for 24 h at 37 °C. Following incubation, growth was measured at a wavelength of 595 nm using a UV-VIS spectrophotometer (Evolution 201, Thermo Fisher Scientific, Waltham, MA), and the fold reduction in erythromycin MIC was deemed to be the modulation factor (Lowrence et al., 2016).

To decipher whether the chosen polyphenol, gallotannin, exhibited synergistic, additive, or antagonistic effect, gallotannin and erythromycin were tested in combination at different concentrations by checkerboard assay against A. baumannii strains (Lowrence et al., 2016). Based on the data, a FIC (fractional inhibitory concentration) index value was calculated; if the FIC values were <0.5, the interaction was deemed synergistic; between 0.5 and 2.0, the interaction was additive; >2.0, the interaction was deemed antagonistic (Odds, 2003).

To evaluate the efflux pump inhibitory potential of the chosen polyphenol, gallotannin, relative to standard inhibitor PAβN, real-time efflux studies were carried out in an XDR clinical isolate of A. baumannii BC2267 strain using ethidium bromide (EtBr) 1 μg/mL as a substrate. In brief, microbial cells were starved, then EtBr was added, followed by supplementation of the cells with 0.4% glucose to activate efflux. Fluorescence due to EtBr accumulation within the cells was measured at Ex = 530 nm and Em = 585 nm (Sundaramoorthy et al., 2018). The enhancement in EtBr fluorescent intensity within the cells due to treatment with gallotannin/standard EPI was deemed as the measure of efflux inhibition activity.

XDR A. baumannii BC2267 strain was grown to 0.4 OD (mid-log phase). The cells were washed and resuspended in PBS and supplemented using 0.4% glucose. EtBr (1 μg/mL) was added to the cells and incubated for 30 min. Gallotannin/standard EPI (PAβN) (16 μg/mL) was added, and subsequently the accumulation of EtBr was immediately measured for the next 20 min with Ex 530 nm and Em 585 nm at 5 min intervals (Sundaramoorthy et al., 2020).

The ability of gallotannin to potentiate the bactericidal activity of erythromycin was evaluated against the XDR clinical isolate A. baumannii BC2267 using a time-kill assay (Grillon et al., 2016). Mid-logarithmic phase cultures were exposed to erythromycin (16 μg/mL) alone or in combination with gallotannin (16 or 32 μg/mL). Erythromycin (16 μg/mL) in combination with PAβN (16 μg/mL) was included as a positive control, while untreated cultures served as the growth control. At specified time points (0, 2, 4, 6, 8, and 24 h), samples were withdrawn, serially diluted, and plated on LB agar. Plates were incubated at 37 °C for 24 h, after which colony-forming units (CFU/mL) were enumerated. The bactericidal activity of the combination treatments was assessed relative to the positive control and individual treatments (Grillon et al., 2016).

1-N-phenylethylamine (NPN) uptake assay was used to assess the ability of gallotannin to permeabilize the outer membrane of XDR A. baumannii (Helander et al., 2000). As the LPS in the outer membrane (OM) of A. baumannii creates steric hindrance to hydrophobic molecules and restricts NPN entry; enhanced NPN fluorescence due to treatment with gallotannin/positive control colistin implies enhanced OM permeability. In brief, exponential-phase cells were collected, washed with 5 mM HEPES buffer containing 0.2% glucose at pH 7.5, and resuspended in an equal volume of the same buffer. NPN was added at a concentration of 0.5 mM, then gallotannin (16 μg/mL)/gallotannin + Erythromycin (16 μg/mL)/gallotannin + colistin was added. NPN fluorescence due to various treatments were measured (Ex = 350 and Em = 420 nm) using a multimode reader (Synergy H1M, Agilent, Santa Clara, CA). NPN in buffer and NPN in buffer along with cells were maintained as controls.

Gallotannin and gallotannin–erythromycin combination were tested for the ability to perturb membrane potential, which was evaluated using the cationic membrane permeabilizing dye DiSC₃(5). Accumulation of the dye in the lipid bilayer of intact cells results in the fluorescence being quenched. When the membrane becomes depolarized, the dye is released into the surrounding aqueous phase, resulting in enhanced fluorescence (Te Winkel et al., 2016). The fluorescence of DiSC₃(5) in buffer was first measured at Ex = 610 ± 5 nm and Em = 660 ± 5 nm. Subsequently, exponential-phase cells were added, which subdued the fluorescence of DiSC₃(5) due to the accumulation of dye within the lipid bilayer of intact cells. Protonophore CCCP was used as the positive control. Gallotannin (16 μg/mL)/CCCP treatments were given, and the resulting variation in fluorescence intensity due to various treatments were quantified using a multimode reader (Synergy H1M, Agilent, Santa Clara, CA).

Reactive oxygen species (ROS) generation in the XDR A. baumannii BC2267 strain following treatment with gallotannin or erythromycin alone, and gallotannin or PAβN in combination with erythromycin, was assessed using the fluorogenic probe 2′,7′-dichlorodihydrofluorescein diacetate (DCFH-DA). In brief, intracellular ROS production was quantified based on the oxidation of DCFH-DA to the fluorescent compound dichlorofluorescein (DCF). Fluorescence intensity was measured using a multimode reader (Synergy H1M, Agilent, Santa Clara, CA) at an excitation wavelength of 485 nm and an emission wavelength of 538 nm (Sundaramoorthy et al., 2019). Hydrogen peroxide was included as a positive control.

Gallotannin toxicity was evaluated in RAW 264.7 macrophages using an MTT assay. RAW macrophages were grown in DMEM with 10% FBS at 37 °C and 5% CO₂. Once they attained 70% confluency, cells were trypsinized, resuspended in fresh media, and ∼10⁶ cells/well were seeded into 96-well plates. Following 24 h incubation, gallotannin at various concentrations (2–48 μg/mL) was added, and cells were incubated further for 24 h; subsequently, MTT (0.5 mg/mL) was added and incubated for 1 h. The formazan crystals formed by the metabolically active cells were dissolved with DMSO, and the absorbance of the extracted fraction was measured at 595 nm to assess cell viability (Peng et al., 2006).

In vivo experiments were conducted in accordance with the guidelines of the Committee for the Purpose of Control and Supervision of Experiments on Animals (CPCSEA; Central Act 26 of 1982). All experimental protocols were reviewed and approved by the Institutional Animal Ethics Committee of SASTRA Deemed University, India (CPCSEA-510/SASTRA/IAEC/RPP). Zebrafish (Danio rerio) were used as the in vivo infection model. Intramuscular infection was established in zebrafish (n = 6 per group) using the XDR A. baumannii clinical isolate BC2267 at an optical density of 0.4, corresponding to approximately 1 × 10⁸ CFU/mL, following Neely et al. (2002) with minor modifications.

At 2 h post-infection, animals were treated with gallotannin or erythromycin alone or with a gallotannin–erythromycin combination, administered as a single intramuscular dose. At 24 h post-treatment, zebrafish were euthanized and decapitated, and the infected muscle tissue was aseptically excised, minced, serially diluted, and plated on LB agar. Following incubation for 24 h, bacterial burden was quantified by enumerating colony-forming units (CFUs). The reduction in bacterial bioburden achieved by gallotannin alone and in combination with erythromycin was determined based on CFU counts and represented graphically.

To quantitatively assess the intracellular accumulation of erythromycin, the antibiotic was derivatized with dansyl chloride, enabling fluorescence-based detection. Dansyl-chloride-tagged erythromycin was used at a final concentration of 16 μg/mL. In brief, exponential-phase cells of A. baumannii BC2267 were harvested, washed with phosphate-buffered saline (PBS), and resuspended in an equal volume of the same buffer. Cell suspensions were then treated with dansyl-tagged erythromycin in combination with gallotannin, PAβN, or colistin for 30 min. Following treatment, cells were washed twice with PBS to remove extracellular antibiotic. The intracellular accumulation of dansyl-conjugated erythromycin under different treatment conditions was quantified by measuring fluorescence at an excitation wavelength of 335 nm and an emission wavelength of 515 nm using a multimode plate reader (Synergy H1M, Agilent, Santa Clara, CA).

All experiments were conducted in triplicate, and statistical analyses were performed using Student’s t-test or one-way ANOVA, as appropriate, with GraphPad Prism version 8.0 (GraphPad Software, La Jolla, CA, United States).

RMSD analysis highlights the overall structural stability of AdeB and its complex with gallotannin during the 100 ns MD simulation (Figure 2A). The unbound AdeB exhibited minimal fluctuations, maintaining a consistent backbone conformation which confirms the intrinsic stability of the protein. Upon gallotannin binding, a gradual increase in RMSD was observed during the initial phase of the simulation, suggesting conformational rearrangements within the binding pocket to accommodate the ligand. After approximately 50 ns, the RMSD reached a plateau, indicating that the system achieved a new equilibrium state. These results suggest that gallotannin binding induces moderate structural flexibility in AdeB without compromising its overall stability, supporting a stable and specific protein–ligand interaction (Baruah et al., 2022).

The RMSF profile reveals that the majority of AdeB residues maintained low atomic fluctuations throughout the simulation, suggesting a stable and rigid core structure (Figure 2B). Localized peaks around residues 15–20 and 60–65 correspond to loop regions or surface-exposed segments that are inherently more flexible. Notably, a pronounced increase in RMSF is observed near the C-terminal region (residues 80–100), indicating significant flexibility in this segment. Such terminal mobility is common and is often associated with regions not directly involved in maintaining structural integrity or ligand binding. Overall, the low RMSF values across most residues confirm that AdeB retains a stable conformation up to 80 residue numbers. The observed fluctuations are confined to the peripheral C-terminal regions (Halder et al., 2023; Nyijime et al., 2025).

Gallotannin exhibited persistent hydrogen bonding with AdeB throughout the molecular dynamics simulation, particularly within the 50–90 ns interval, where it consistently maintained at least four hydrogen bonds, indicative of steady interaction stability (Figure 2C). Notably, the highest number of hydrogen bonds observed was seven, occurring around the 96th nanosecond, while intermittent fluctuations in hydrogen bond count reflect inherent ligand–protein flexibility. Overall, these results demonstrate that gallotannin sustained stable binding with AdeB across the simulation, underscoring strong affinity and dynamic adaptability in the ligand–protein interface (Halder et al., 2023).

The ability of 13 polyphenols (quercetin, shikimic acid, piperine, myricetin, gallic acid, quinic acid, gallotannin, kaempferol, syringic acid, naringenin, caffeic acid, naringin, and picroside) to reverse/reduce the MIC of erythromycin was evaluated against a reference strain of A. baumannii (MTCC1425) and an XDR clinical isolate (BC2267). Among the evaluated polyphenols, only gallotannin (1,2,6-tri-O-galloyl-bD-glucopyranose) caused a 64-fold MIC reversal of erythromycin against both isolates (Supplementary Table S5); hence, it was chosen for further studies. Subsequently, gallotannin was tested to reverse the MIC of antibiotics belonging to three different classes (cell wall inhibitor—meropenem, replication inhibitor—ciprofloxacin, and protein synthesis inhibitor—erythromycin) against five different clinical strains of A. baumannii and one reference strain, A. baumannii (MTCC1425). Gallotannin successfully reduced the MIC of all three antimicrobials against all the isolates of A. baumannii ∼64-fold (Table 3). In addition, we tested the ability of PAβN (standard efflux pump inhibitor) to reverse the MIC of these antibiotics against all six strains. Our observations showed that PAβN exerted significant reversal erythromycin MIC in the reference strain A. baumannii (MTCC1425) and two clinical isolates (BC2267& E1406), but for the other two antibiotics—meropenem and ciprofloxacin—PAβN failed to significantly (>2-fold) reverse the MIC of tested antibiotics against clinical isolates (Supplementary Table S6). As A. baumannii exhibits other mechanisms of resistance, such as OXA type Class D β-lactamase-mediated resistance for meropenem and gyrA/parC target site mutation mediated resistance for ciprofloxacin, efflux inhibition by PAβN (standard inhibitor) is unable to restore drug sensitivity in these isolates. It is noteworthy that gallotannin caused >64-fold MIC reversal for both meropenem and ciprofloxacin against five different clinical isolates (Table 3), implying that gallotannin apart from its efflux inhibitory potential also exhibits meropenem/ciprofloxacin resistance modulatory potential, which remains to be explored in further studies. Especially with ciprofloxacin, gallotannin fully restored the sensitivity of the fluoroquinolone in the XDR strain BC2267, MDR strains U3154 and 1,406, and the reference strain AB1425. Thus, by its synergistic interaction, Gallotannin succeeded in increasing the intracellular concentration of different antibiotics in five clinical isolates of A. baumannii, highlighting its strong efflux inhibitory potential. Erythromycin was chosen for further study since it is a less permeable macrolide antibiotic and previous studies have shown that erythromycin also exhibits biofilm inhibitory potential (Dong et al., 2024).

Among the clinical isolates of A. baumannii tested, since BC2267 was the sole XDR strain that displayed high levels of resistance to all the antimicrobials tested, further studies were carried out using it. Synergistic interaction between polyphenols and erythromycin was evaluated against the XDR BC2267 strain and among the polyphenols evaluated, and only gallotannin displayed synergy with the BC2267 strain with an FIC index of 0.125.

To evaluate whether the MIC reversal caused by gallotannin is due to efflux inhibition, real-time efflux (RTE) and time-dependent accumulation (TDA) studies were performed. EtBr is a common substrate for most efflux transporters, and a compound inhibiting efflux will result in the enhanced accumulation of EtBr within the bacterial cells, which can be quantified using RTE/TDA. A real-time efflux assay was performed for the XDR BC2267 strain. Cells were initially starved by incubation in the absence of glucose. Subsequently, these cells were subjected to gallotannin/PaβN/Verapamil/PBS(Control) treatment for 1 h. Following this, the cells were supplemented with glucose, and fluorescence due to EtBr was measured for 20 min. RTE results revealed that gallotannin caused the enhanced inhibition of EtBr efflux relative to the positive controls PAβN (Standard RND pump inhibitor) (Nikaido and Pagès, 2012) and verapamil, a clinically approved calcium channel blocker used in the treatment of hypertension, which is widely employed as a standard inhibitor of MATE-family efflux pumps (Radchenko et al., 2015) (Supplementary Figure S4).

TDA focuses on net intracellular accumulation of EtBr within the bacterial cells. Cells supplemented with glucose tend to pump EtBr out through its efflux transporters. Efflux inhibition by gallotannin/standard inhibitors (PAβN and verapamil) leads to increased EtBr accumulation over time. The results (Figure 3) show that gallotannin causes a slight but enhanced accumulation of EtBr within XDR A. baumannii cells relative to positive control PAβN, which further corroborates the observation that gallotannin reverses the MIC of three different antimicrobials against five different clinical isolates of A. baumannii (Table 3).

A time-kill assay was performed to evaluate the ability of gallotannin to potentiate the bactericidal effect of erythromycin. The concentrations of gallotannin and erythromycin employed for the time-kill assay were determined based on a checkerboard assay, which showed that the minimum effective concentration of gallotannin and erythromycin was 16 μg/mL. For the time-kill assay, mid-log cells of XDR A. baumannii BC2267 corresponding to 0.4 OD were subjected to the following treatments: gallotannin alone, erythromycin alone, gallotannin (16 μg/mL) + erythromycin (16 μg/mL), and gallotannin (32 μg/mL) + erythromycin (16 μg/mL); PAβN (16 μg/mL) + erythromycin (16 μg/mL) served as the positive control and the untreated culture served as the growth control. The plate counts at different time intervals revealed that treatment with gallotannin/erythromycin individually resulted in a trend similar to the untreated control, with a 2–3 log increase in cell counts by 24 h (Figure 4). Relative to treatment with erythromycin alone, a combination of gallotannin (16 μg/mL) with erythromycin resulted in ∼2 log decline by 24 h, which was similar to the reduction in cell counts induced by PAβN + erythromycin. However, a combination of 32 μg/mL of gallotannin with erythromycin (16 μg/mL) resulted in a significant 3 log decline, implying that gallotannin in combination with erythromycin enhanced the antimicrobial potential of erythromycin and effectively curtailed the growth of the XDR A. baumannii BC2267 strain.

As CRAB is a pressing issue with respect to nosocomial infections, we evaluated the ability of gallotannin to potentiate the bactericidal effect of meropenem (Carbapenem) in XDR A. baumannii strain using a time-kill assay (Supplementary Figure S5). Interestingly, meropenem treatment alone, like erythromycin, prevented the growth of A. baumannii but was unable to significantly reduce the founder population, whereas a gallotannin–meropenem combination caused a significant 2 log decline in cell counts, indicating that gallotannin by virtue of its efflux inhibitory potential causes increased the accumulation of meropenem, resulting in an enhanced bactericidal effect. These observations imply that a similar transporter could be involved in extruding both meropenem and erythromycin, the nature of which remains to be explored in future studies.

Gallotannin at various concentrations (2–48 μg/mL) was tested for its toxicity in raw macrophage cell lines. Approximately 10⁶ cells were exposed to different concentrations of gallotannin in triplicate, and after 24 h of incubation, cell viability due to treatment was assessed by MTT assay. The results (Supplementary Figure S6) revealed that gallotannin did not significantly impact the viability of macrophages; 80%–90% viability was retained at the different concentrations tested (2–32 μg/mL), and at 48 μg/mL, 70% viability was maintained. This indicates that gallotannin is relatively non-toxic to cultured raw macrophages, and at the concentration (16 μg/mL) employed throughout the present study, cells maintained ∼85–90% viability.

To evaluate the ability of gallotannin to potentiate the antimicrobial effect of erythromycin against the XDR A. baumannii BC2267 strain, an in vivo zebrafish infection study was undertaken. Adult fish (n = 6) were independently infected with 0.4 OD of A. baumannii cells, corresponding to ∼1 × 10⁸ CFU/mL 2 h post-infection, and were treated with erythromycin alone, gallotannin alone, or a combination of erythromycin and gallotannin; a PAβN and erythromycin combination was used as positive control. Fish were euthanized 24–48 h post-infection, and muscle tissue was dissected, macerated, serial diluted, plated on LA plates, and incubated for 24 h at 37 ^°C. Plate counts (Figure 5) revealed that the untreated control displayed a colony count of ∼9 log CFU. The gallotannin-treated group displayed ∼8 log CFU. Erythromycin treatment also resulted in colony counts comparable to gallotannin treatment. The positive control PAβN + erythromycin combination resulted in a minimal decline in cell counts ∼ 1 log decline in CFU relative to the untreated control. Among all treatments, gallotannin + erythromycin caused a significant 4 log decline in cell counts relative to the untreated control (Figure 5), underscoring the ability of gallotannin to potentiate the bactericidal effect of erythromycin in vivo. Despite the effectiveness of PAβN + erythromycin treatment in the time-kill assay, PAβN failed to potentiate the bactericidal effect of erythromycin in the zebrafish infection model for reasons unknown.

To gain mechanistic insights into the mode of action of gallotannin in synergizing the antibacterial potential of erythromycin, various assays like membrane permeability, intracellular accumulation using fluorescently tagged erythromycin, membrane potential, and ROS generation were undertaken.

Membrane permeability was determined using the fluorophore NPN. The results of the membrane permeability assay (Figure 6) revealed that untreated cells displayed baseline membrane permeability. Treatment with gallotannin caused a 6-fold increase in membrane permeability, similar to that caused by erythromycin. Positive control colistin induced an 8-fold increase in membrane permeability relative to untreated cells. However, treatment with a combination of gallotannin + erythromycin resulted in ∼10-fold increase in membrane permeability, indicating that gallotannin + erythromycin effectively permeabilizes the inner membrane of XDR A. baumannii, which also explains why gallotannin is effective in potentiating the bactericidal effect of erythromycin.

In order to determine whether gallotannin causes the accumulation of erythromycin within the bacterial cells, erythromycin was coupled with dansyl chloride, incubated with A. baumannii cells, and co-treated with either gallotannin/PAβN/colistin for 30 min and followed by washing. The accumulation of dansyl-tagged erythromycin within the cells was quantified using a spectrofluorimeter. Cells treated with only dansyl-tagged erythromycin were maintained as a control. The results revealed that positive control colistin, which permeabilizes the membrane to facilitate erythromycin entry, caused ∼2-fold increase in fluorescence relative to the untreated control (Figure 7), whereas gallotannin caused a significant ∼3-fold increase in fluorescence due to the accumulation of tagged erythromycin within the cells (Figure 7). Interestingly, PAβN caused only a slight increase in tagged erythromycin accumulation relative to the untreated cells, which is not statistically significant. Thus, our observations indicate that gallotannin is actually effective in causing a significant accumulation of erythromycin within the cells, which could account for enhanced bactericidal effect of gallotannin-erythromycin combination.

Many compounds exert an antibacterial effect by inducing ROS. Earlier studies have shown that, at least under aerobic conditions, conventional antibiotics induce ROS (Van Acker and Coenye, 2017). We were interested in exploring whether gallotannin exerts additional functional roles such as generating ROS apart from inhibiting drug efflux in A. baumannii. Our observations based on the ROS assay reveal that gallotannin caused ∼1.25-fold increase in ROS relative to untreated cells. As erythromycin is impermeable in Gram negatives, it did not alter ROS levels relative to untreated cells. When a combination of erythromycin and gallotannin was employed, a significant 1.5–1.8-fold increase in ROS was observed, which was greater than the ROS generation observed in cells treated with hydrogen peroxide (Figure 8). The observed enhancement in ROS could be attributed to two features of gallotannin: i) enhanced permeability allowing better access of erythromycin within bacterial cells and ii) efflux inhibition, which prevents the extrusion of accumulated erythromycin. Both these features generated sufficient ROS that could possibly attribute for the enhanced bactericidal effect of the gallotannin + erythromycin combination (Figures 4, 5). Positive control PAβN + erythromycin induced ROS comparable to H₂O₂ treatment but was unable to induce elevated ROS similar to gallotannin + erythromycin.

A membrane potential assay was performed to evaluate whether gallotannin dissipated the proton motive force (PMF), which energizes proton-based efflux transporters. Protonophore CCCP was used as the positive control and DiSC₃(5) as the probe. Membrane depolarization will result in removal of the dye from the lipid bilayer, leading to enhanced fluorescence. Treatment with gallotannin did not result in increased fluorescence, implying that, unlike protonophore CCCP, gallotannin did not cause membrane depolarization and probably does not inhibit efflux pumps by cutting off the energy source, like protonophore CCCP (Supplementary Figure S6).