Screening of the Pandemic Response Box library identified promising compound candidate drug combinations against extensively drug-resistant Acinetobacter baumannii

Infections caused by antimicrobial-resistant Acinetobacter baumannii pose a significant threat to human health, particularly in the context of hospital-acquired infections. As existing antibiotics lose efficacy against Acinetobacter isolates, there is an urgent need for the development of novel antimicrobial agents. In this study, we assessed 400 structurally diverse compounds from the Medicines for Malaria Pandemic Response Box for their activity against two clinical isolates of A. baumannii: A. baumannii 5075, known for its extensive drug resistance, and A. baumannii QS17-1084, obtained from an infected wound in a Thai patient. Among the compounds tested, seven from the Pathogen box exhibited inhibitory effects on the in vitro growth of A. baumannii isolates, with IC50s ≤ 48 µM for A. baumannii QS17-1084 and IC50s ≤ 17 µM for A. baumannii 5075. Notably, two of these compounds, MUT056399 and MMV1580854, shared chemical scaffolds resembling triclosan. Further investigations involving drug combinations identified five synergistic drug combinations, suggesting potential avenues for therapeutic development. The combination of MUT056399 and brilacidin against A. baumannii QS17-1084 and that of MUT056399 and eravacycline against A. baumannii 5075 showed bactericidal activity. These combinations significantly inhibited biofilm formation produced by both A. baumannii strains. Our findings highlight the drug combinations as promising candidates for further evaluation in murine wound infection models against multidrug-resistant A. baumannii. These compounds hold potential for addressing the critical need for effective antibiotics in the face of rising antimicrobial resistance.


Bacterial strains, culture conditions, and whole-genome sequencing
Extensively drug-resistant A. baumannii strains, namely QS17-1084 and 5075, were employed in this study.QS17-1084 was isolated from a wound infection of a patient at Queen Sirikit Naval Hospital, while A. baumannii 5075 was obtained from the Walter Reed Army Institute of Research, USA 16,17 .All A. baumannii strains were cultured in cation-adjusted Mueller Hinton broth (CAMHB) (Becton, Dickinson, USA).Whole-genome sequencing was performed using MiSeq or NextSeq benchtop sequencers (Illumina, Inc., Diego,CA) at the Multidrug-resistant Organism Repository and Surveillance Network (MRSN) 18 .The genome data of A. baumannii strains 5075 were deposited in GenBank under the accession number AHAH00000000 16 .The genomic sequence of QS17-1084 is available in NCBI database (BioProject No. PRJNA814829, BioSample No. SAMN42720897 and SRA No. SRR29910022) 17 .

Pandemic Response Box compounds screening against A. baumannii
The initial 10 mM stock plates of the 400 compounds were diluted in 100% DMSO to a concentration of 1 mM.The bacterial growth inhibition assay followed Clinical and Laboratory Standards Institute (CLSI) microdilution method guidelines 19,20 .Bacterial colonies were incubated with 10 µM of each compound in cation-adjusted MH broth, and optical density at 600 nm (OD600) was measured.Percent growth of compound-treated A. baumannii was calculated using the following formula: Compounds inhibiting bacterial growth by more than 50% were selected for further testing [Results section specifies the compounds].

Structural similarity measurements
Structural similarity between Pandemic Response Box compound hits and commonly used antibacterial drugs was calculated using atom pair fingerprints (APfp) 21 .APfp values were determined and subjected to hierarchical clustering analysis using ChemMineR software 22,23 .

Antimicrobial susceptibility testing of the selected compounds
Antimicrobial susceptibility testing (AST) of A. baumannii QS17-1084 and 5075 against known antibiotic panels was performed using the Phoenix platform (panel NMIC/ID504; BD Diagnostics, NJ, USA).We determined the minimal inhibitory concentrations (MICs) and minimal bactericidal concentrations (MBCs) of 7 selected MMV Pandemic response box compounds, doxycycline hydrochloride (WRAIR), and imipenem (US Pharmacopeia, Frederick, USA) were determined in triplicate using broth microdilution (BMD) using Clinical and Laboratory OD600 of compound treated A. baumannii − OD600 of the blank well) × 100 (OD600 of DMSO treated A. baumannii − OD600 of the blank well) Standards Institute (CLSI) guidelines 21 .Additionally, methods were performed in accordance with the relevant guidelines and regulations of the Armed Forces Medical Research Institute of Medical Sciences Scientific Review Committee and Institutional Review Board.For MICs determination, the compounds were diluted as twofold serial dilutions 20 .Bacteria were suspended in a clear glass tube containing with CAMHB before adjusting the inoculum suspension to 0.5 MacFarland standard using densitometer (Den-1, Biosan, USA).Subsequently, the inoculum suspension was prepared to a density yielding of the final bacteria concentration approximately 5 × 10 5 CFU/ml.Twenty microliters of the inoculum were subjected to 96 well microtiter plates (Thermo Scientific, MA, USA) with 20 μL of a series of drug-like compounds and control antimicrobial agents, to make up the final volume of 40 μL (doxycycline and imipenem) concentrations, except the last well that was left for media only.Doxycycline and imipenem were also used as quality control drugs for E. coli ATCC 25,922 and P. aeruginosa ATCC 27853, respectively.A. baumannii incubation conditions were 35 ± 2 °C for 20-24 h while 16-20 h for E. coli and P. aeruginosa.All experimental sets were performed in triplicate.
For MBCs determination, the solution of the last clear MIC well of each selected compound was further incubated onto sheep blood agar (Becton, Dickinson, USA) and MacConkey agar (Becton, Dickinson, USA) before incubating at 37 °C in ambient air for 20-24 h.MBCs was determined by observing the bacterial growth on the sheep blood agar and MacConkey agar plates.
For IC 50 (50% inhibitory concentration) determination, A. baumannii strains were treated with various concentrations of test compound and cultured according to the microdilution method above.IC 50 s were estimated by nonlinear regression analysis using GraphPad Prism (San Diego, CA, USA).

Drug combinations
Fixed ratio combinations of various antimalarial drugs were performed as previously described with some modifications [24][25][26] .Stock solutions of the selected Pandemic response box compounds as well as control drugs doxycycline (DOX) (WRAIR), erythromycin (ERY) (Sigma-Aldrich, Missouri), ciprofloxacin (CIP) (Sigma-Aldrich, Missouri), and 7 compounds from the MMV Pandemic Response Box, according to their %inhibition, were prepared as follows: 2 mM and 4 mM DOX in 70% ethanol for A. baumannii 5075 and QS17-1084, respectively; 10 mM ERY in 70% ethanol, 10 mM CIP in 0.1 N HCl, and 5 mM MMV compound in DMSO.The total of 64 drug combinations were performed.The solutions for each drug were combined in ratios of 1 + 1, 1 + 3, 3 + 1, 1 + 4, 4 + 1, and 1 + 2; with each drug also tested alone.20 µL of single and combination drug solutions were then introduced into 96-well plates to give a row with two-fold serial dilutions.20 µL of A. baumannii with a final 5 × 10 4 CFU per well were added, and the test plates were incubated for 20-24 h at 37 °C.Bacteria growth was measured by OD 600 as described above and the individual 50% fractional inhibitory concentrations (FIC 50 ) were determined as previously described 27 .Isobolograms were constructed by plotting the FIC 50 of drug A against the FIC 50 of drug B for each of the six drug ratios.A concave curve indicated synergy, a straight-line represented additivity and a convex curve indicated antagonism.To obtain numeric values for the interaction, results were expressed as the sum of the FIC 50A and FIC 50B .The sum FIC 50 (ΣFIC 50 ) values indicate the kinds of interaction as follows: synergism when ΣFIC 50 < 0.5; toward synergism when ΣFIC 50 < 1; additive when ΣFIC 50 = 1; toward antagonism when ΣFIC 50 > 1; antagonism when ΣFIC 50 > 2 to 4. The IC 50 s of each drug in the test combination were standardized by allocating the value of 1 to each drug that was tested alone and prorated values for each fixed concentration ratio.

Time-kill assay
In vitro time-kill assay was performed to evaluate antibiotic efficacy of drug combinations compared to individual compounds.According to drug the combination assay, MUT056399, brilacidin, and their combinations were tested against A. baumannii QS17-1084, while MUT056399, eravacycline, and their combinations were tested against A. baumannii 5075.Each drug was prepared separately to the desired concentration based on MIC determined against each A. baumannii strain.In brief, equal volumes (100 µL) of drug alone or drug mixture and bacterial suspension were added to a sterile 96-well flat bottom plate (Corning, NY, USA).CAMHB medium was used to dilute drug(s) and bacteria.To prepare bacterial inoculum, bacterial culture was adjusted to 0.5 McFarland turbidity and diluted to 1:100 then inoculated into the wells containing drugs in duplicates.Plates were incubated in a 35 ± 2 °C incubator.Bacterial growth was observed by CFU count and OD 600 measurement for 6 time points at 0, 2, 4, 6, 8, and 24 h incubation time.CFU count was performed by spreading the 10-folds serially diluted suspension onto tryptic soy agar plates.Results were expressed in CFU/mL.The limit of detection was set at 5 CFU/mL.The reduction in CFU/mL at 24 h was used to interpret results following the criteria; the reduction of ≥ 2 log 10 , but < 3 log 10 was defined as bacteriostatic activities, and the reduction of ≥ 3 log 10 was defined as bactericidal activities, relative to initial inoculum.For drug combination, the reduction of ≥ 2 log 10 relative to the most active component alone was defined as synergistic activities 28 .Experiments were performed in two independent runs.

Inhibition of biofilm formation
The effect of synergistic drug combinations were tested on the inhibition of biofilm formation of A. baumannii QS17-1084 and A. baumannii 5075.The concentrations of drugs were monitored to observe synergistic effects using a time-kill assay.Briefly, bacterial culture was adjusted to 0.5 McFarland turbidity in tryptic soy broth (TSB).Drugs were diluted in TSB with 1% glucose.Equal volumes (100 µL) of the 0.5 McFarland bacterial suspension and drug combination were added to a 96-well polystyrene non-treated, flat-bottom plate.This obtained a final concentration of 0.5% glucose in TSB.The respective strain of A. baumannii (without drugs) was used as a growth control and medium alone was used as a blank.In each run, A. baumannii ATCC19606 and E. coli ATCC25922 were included as positive and negative biofilm producing strains, respectively.Plates were incubated at 37 °C for 24 h with static condition.After incubation, the plates were gently washed three times with 300 µL of phosphate-buffered saline to remove planktonic and loosely adhered bacteria.The plates were fixed with 200 µL of methanol for 15 min, air dryed for 30 min, and stained with 175 µL of 0.5% crystal violet for 10 min.The plates were washed three times with 300 µL of distilled water and the crystal violet stain was solubilized by adding 175 µL of 33% acetic acid.After shaking for 10 min at room temperature, the solution was transferred to another plate and measured at OD 595 .Experiments were performed in three independent runs.
For result interpretation, the optical density cut-off (ODc) was defined as 3 standard deviation (SD) above mean OD of negative control.The degrees of biofilm formation were classified as follows: strong biofilm formation (OD > 4 × ODc), moderate biofilm formation (2 X ODc < OD ≤ 4 × ODc), weak biofilm formation (ODc < OD ≤ 2 × ODc), and non-biofilm formation (OD ≤ ODc).To calculate inhibition of biofilm formation, the absorbance values for the blank wells were subtracted from the values for the test wells to minimize background interference.Percentage of inhibition of biofilm formation was calculated as follows.

Statistical analysis
Statistical analysis was performed using GraphPad Prism 10.0 (GraphPad Software, Inc., San Diego, CA, USA).The difference of data between groups was assessed by multiple t test, unless otherwise stated.Statistical significance was defined as a P value < 0.05.

Phenotypic identification
Initially, we selected A. baumannii QS17-1084 since we are interested in using this strain of bacteria further in our animal wound model as part the WRAIR/AFRIMS drug development pipeline.A. baumannii 5075 has been previously employed in animal models 29,30 .A. baumannii QS17-1084, collected from the infected wound of a patient, was phenotypically identified as well as their antimicrobial susceptibility, in comparison to A. baumannii 5075, using the Phoenix M50 automated microbiology platform with NMIC/ID504 panel.Antimicrobial susceptibility testing (AST) results revealed A. baumannii QS17-1084 was resistance to almost all antimicrobial classes including aminoglycosides (amikacin and gentamicin), carbapenems (imipenem and meropenem), cephalosporins (ceftazidime, and cefepime), ciprofloxacin, piperacillin-tazobactam, and trimetroprim-sulfamethoxazole (Fig. 1), similar to A. baumannii 5075, except for tetracyclines (minocycline) in which A. baumannii QS17-1084 was intermediate while A. baumannii 5075 susceptible.In addition, both A. baumannii strains presented the MIC of 2 µg/ml towards colistin, indicative of colistin intermediate.

Primary screening
Initially, a primary screen concentration of 10 µM of the Pandemic response box compounds was carried out against A. baumannii QS17-1084, using the CLSI microdilution method.As part of our target product profile compound with > 80% inhibitory effect are advanced for further investigation.The screening assay produced 7 compounds that had a > 80% inhibitory effect against both A. baumannii QS17-1084 and 5075 (Fig. 2 and Table 1).Five of the seven compounds with inhibitory activities including alexidine, gepotidacin, eravacycline, epetraborole, and brilacidin are reference compounds with known antifungal and antibacterial activity.Another two compounds, MUT056399 and MMV1580854, sharing a core structure of diphenyl ethers, inhibited 95% of the growth of A. baumannii 5075 while inhibited A. baumannii QS17-1084 to around 80-86% in the primary screen.All seven compounds inhibited the growth of A. baumannii QS17-1084 strain greater than erythromycin, doxycycline, imipenem, and ciprofloxacin (Fig. 2).

Hierarchical clustering analysis
Next we performed the hierarchical clustering analysis (HCA) of the 7 compounds and 4 commonly known antibacterial drugs, including ciprofloxacin, tetracycline, colistin and gentamicin (Fig. S1).The HCA revealed that gepotidacin, a first-in-class triazaacenaphthylene antibiotic inhibiting bacterial DNA replication, showed the maximum structural similarity with ciprofloxacin, a fluoroquinolone antimicrobial agent.Eravacycline, a synthetic halogenated tetracycline displayed close structural similarity with tetracycline, while MUT056399 was structurally closed to MMV1580854, suggesting that they may inhibit the same biological target, which is enoyl acyl carrier protein reductase (FabI).Epetraborole, a class of boron-heterocyclic antimicrobials targeting leucyl-tRNA synthetase (LeuRS) 32  exhibits the maximum structurally similarity with colistin (also known as polymyxin E) and disrupts bacterial cell membranes.

Evaluation of antibacterial effects of the selected drugs on A. baumannii
To further understand the effect of screening drugs on A. baumannii, the MIC and MBC assays were utlized (Table 2).For A. baumannii QS17-1084, alexidine and brilacidin exhibited the lowest MIC values of 10 µM, while eravacycline and epetraborole provided the same MIC values of 20 µM.Gepotidacin and imipenem showed Vol.:(0123456789)To better understand drug sensitivity, we subsequently determined the IC 50 values, the concentration of drug required for 50% inhibition, of 9 compounds against both A. baumannii QS17-1084 and 5075 (Fig. 3A).For A. baumannii QS17-1084, 3 compounds, including gepotidacin, eravacycline, and epetraborole, showed a relatively good drug sensitivity with the IC 50 values ranging from 1.61 to 1.99 µM.Four compounds including alexidine, MMV1580854, brilacidin, and imipenem provided a comparable IC 50 values ranging from 4.22 to 7.17 µM, while MUT056399 and doxycycline had the elevated IC 50 values of 12.14 and 84.87 µM, respectively.When the compounds were tested against A. baumannii 5075, the IC 50 values of 5 compounds, including MMV1580854, brilacidin, MUT056399, doxycycline, and imipenem, were altered significantly compared with A. baumannii QS17-1084 (Fig. 3A), while those of alexidine, gepotidacin, eravacycline, epetraborole were not significantly changed.Of five compounds, MMV1580854, MUT056399 and doxycycline showed an increase in drug sensitivity towards A. baumannii 5075.Doxycycline showed the 314-fold reduced IC 50 value, followed by MUT056399 (8.20-fold decrease), and MMV1580854 (2.71-fold decrease).Two compounds, brilacidin and imipenem, exhibited the increased IC 50 values against A. baumannii 5075 around 1.94-and 1.52-fold increases, respectively.In addition to IC 50 values, IC 99 values of each compound were also determined (Table S1).The IC 99 concentration which is approximately 100-fold the IC 50 concentration is used when complete inhibition is required.The IC 99 values of some compounds against A. baumannii QS17-1084 were comparable to MIC values, including alexidine, brilacidin, while for the rest of the compounds, the IC 99 values were smaller than the MIC values.In the case of A. baumannii 5075, the IC 99 values of eravacycline, epetraborole, and doxycycline were close to the MIC values.While the IC 99 values of MMV1580854, MUT056399, and imipenem were less than the MIC values, those of alexidine, gepotidacin and brilacidin were higher than the MIC values in A. baumannii 5075.Since MMV1580854 shares chemical structure similarity with MUT056399, we closely compared their IC 50 values (Fig. 3 B).When both compounds were tested against A. baumannii 5075, the IC 50 values of both compounds were not significantly different (1.51 µM for MMV1580854 and 1.49 µM for MUT056399.However, when tested against A. baumannii QS17-1084, the IC 50 value of MMV1580854 was 2.87-fold decreased compared to the IC 50 value of MUT056399, indicating that MMV1580854 was more potent than MUT056399 when treated against A. baumannii QS17-1084.

Drug combination effects on A. baumannii
Finding novel antimicrobial agents and therapies based on synergistic combinations are crucial to battle drugresistant bacteria.Of 7 potent compounds from the Pandemic Response Box, and 3 known antibiotics, the 46 drug combinations were investigated (Fig. 4) to evaluate the synergistic effects using previously published methods 25 .Two drug combinations (brilacidin-MMV1580854, and brilacidin-MUT056399) showed toward synergistic effects (ΣFIC 50 < 1) on A. baumannii QS17-1084, while 3 combinations (eravacycline-MMV1580854, eravacycline-brilacidin, and eravacycline-MUT056399) had toward synergistic activity against A. baumannii 5075.It is interesting that out of the 6 combinations with synergism, they were stemmed from 4 compounds including brilacidin, eravacycline, MMV1580854, and MUT056399, the latter two of which shared similar chemical scaffold.Antagonistic effects (ΣFIC 50 > 1) were also observed in 5 drug combinations for A. baumannii QS17-1084 and 4   drug combinations for A. baumannii 5075.Three drug combinations showing antagonistic interactions in both A. baumannii strains were ciprofloxacin-erythromycin, gepotidacin-alexidine, and eravacycline-erythromycin.

Inhibition of biofilm formation
The ability of drug combinations in inhibiting biofilm formation of A. baumannii QS17-1084 and A. baumannii 5075 was observed using crystal violet staining.Generally, both A. baumannii QS17-1084 and A. baumannii 5075 could produce biofilms but less than a positive control A. baumannii ATCC19606 (Fig. 6).It was revealed that the combination of MUT056399 and brilacidin (ratio 1:2) significantly inhibited biofilm formation of A. baumannii QS17-1084 and A. baumannii 5075, respectively in relative to bacteria alone without drug treatment (Fig. 6A and B).Percent inhibition of biofilm formation at 24 h was shown 80% and 85% when treated A. baumannii QS17-1084 and A. baumannii 5075, respectively with drug combination.Further, biofilm formation of both strains was profoundly inhibited after 24 h drug exposure, which the detected biofilms were presented in a similar manner to negative control strain, E. coli ATCC25922 (Fig. 6A and 6B).

Discussion
The purpose of this work is to find inhibitors against extensively drug-resistant A. baumannii, which is a prominent member of ESKAPEE group of pathogens, and primarily associated with wound and burn infections and ventilator-associated pneumonia 33 .The 400 MMV Pandemic Response Box compounds were screened, and 7 compounds were selected and further evaluated through phenotypic experiments against 2 extensively Fig. 6.Effects of drug combination on preventing biofilm formation.(A) A. baumannii QS17-1084 was treated with MUT056399 and brilacidin at the ratio 1:2.(B) A. baumannii 5075 was incubated with MUT056399 and eravacycline at ratio 1:2.The drug concentrations were similar to those used in the time-kill assay.A. baumannii ATCC19606 and E. coli ATCC25922 were included as positive and negative biofilm producing strains, respectively.P < 0.05 was analyzed by using Student's t-test.The representative images are biofilms stained by crystal violet.
drug-resistant A. baumannii strains, including QS17-1084 and 5075.In addition, we showed that the combinations of brilacidin and diphenyl ether derivatives, and eravacycline and brilacidin or diphenyl ether derivatives showed synergistic effects on A. baumannii QS1084 and 5075, respectively.
A. baumannii remains a heterogeneous species with over 1,380 different sequence types (STs) identified by traditional Pasteur scheme multilocus sequence typing (MLST) 34,35 .A. baumannii 5075 and QS17-1084 belonged to ST1 and ST2, respectively.By comparing with A. baumannii 5075, A. baumannii QS17-1084 had ant(3″)-IIa, aph(3″)-Ib, aph(6)-Id genes that encode aminoglycoside adenylyltransferase and aminoglycoside phosphotransferase, and armA gene that encodes 16S rRNA methyltransferase conferring high level resistance of various aminoglycosides 36 .The presence of these genes is well correlated with the drug susceptibility test in which both A. baumannii QS17-1084 and 5075 are resisted to amikacin and gentamicin (Fig. 1). A. baumannii QS17-1084 may resist to β-lactams via Class C and D β -lactamases as well as altering penicillin-binding proteins (PBP3) while A. baumannii 5075 may employ Class A, C and D β-lactamases for β-lactams resistance.The bla ADC belongs to class C β-lactamases.The ADC β-lactamase are cephalosporinase with resistance extended-spectrum cephalosporins 37 .The bla oxa genes, classified for class D β-lactamases resistance, were reported in most of carbapenemase producing A. baumannii in several studies in Southeast Asia, including Thailand 38 , Vietnam 39 , Malaysia 40 , China 41 as well as in Europe 42 .The bla oxa-66 was grouped into intrinsic bla oxa-51-like oxacillinase gene found in A. baumannii 37 .Both A. baumannii strains in this study showed resistance to ciprofloxacin because they harbored the mutations on gyrA and parC genes.The most distinction of these two A. baumannii strains is that A. baumannii QS17-1084 contained genes that confer resistance to macrolides and tetracyclines, so it showed intermediate against minocycline (Fig. 1) and elevated MIC and IC 50 values (Table 2 and Fig. 3).
The screening of 400 diverse compounds identified seven compounds (Table 1 and Fig. 2), including alexidine, gepotidacin, MMV1580854, eravacycline, epetraborole, brilacidin, MUT056399.Alexidine is an anticancer drug, targeting a mitochondrial tyrosine phosphatase, PTPMT1, in mammalian cells and causing mitochondrial apoptosis 43 .It also inhibited planktonic growth, prevented biofilm formation as well as killing biofilms formed by diverse fungal organisms 44 .Alexidine dihydrochoride was also identified in a quantitative high throughput screen (qHTS) against A. baumannii 5075 with the IC 50 of 29.02 µM.Gepotidacin is a novel triazaacenaphtylene antibiotic with inhibiting DNA gyrase and topoisomerase IV.Its activity was shown against most strains of E. coli and Staphylococcus saprophyticus 45 .MMV1580854 and MUT056399, which are diphenyl ether derivatives, have been shown as inhibitors against enoyl-acyl carrier protein reductase (FabI) 46,47 .The IC 50 values of MMV1580854 and MUT056399 againstS.aureus FabI was 56 and 12 nM, respectively and the MIC value of MMV1580854 against S. aureus CIP54,14 was about eightfold higher than that of MUT056399 47 .In our study, while the IC 50 values of both MMV1580854 and MUT056399 against A. baumannii 5075 were very close (Table S1), MMV1580854 seemed to be more potent than MUT056399 against A. baumannii QS17-1084.Eravacycline, a newly developed, fully synthetic tetracycline derivative with broad-spectrum activity against extended spectrum β-lactamase producing Enterobacteriaceae and Acinetobacter 48 , inhibits bacterial protein synthesis through binding to the 30S ribosomal subunit.Eravacycline exhibited greater activity than the comparators of the tetracycline class with the MIC 50 of 0.5 mg/L against multidrug-resistant, carbapenem non-susceptibility A. baumannii isolates. 49.Since there is no information on clinical breakpoints against A. baumannii, Deolankar et al. suggested that the MIC values (< 4 μg/ mL) of eravacyclines were effective against their 19 A. baumannii patient isolates 50 .In this study, the MIC value of eravacycline against A. baumannii 5075 was 1.25 μM (or 0.79 μg/mL), while that against A. baumannii QS17-1084 was 20 μM (or 12.6 μg/mL), which may be accounted for the presence of tet(B) gene.Cheng et al. screened 25 antibiotics commonly used for infections by gram negative bacteria and found that nine compounds could inhibit the growth of A. baumannii 5075 in concentration dependence (IC 50 ranging from 0.39 to 66.5 µM) 51 .Tigecycline, a derivative of the tetracycline, was the most potent with the IC 50 of 0.15 µM.This was in accordance with our result that eravacycline was the most potent with the IC 50 of 0.22 µM 51 .Epetraborole, a new class of leucyl-tRNA synthetase inhibitors, showed potent activity against a broad range of Gram-negative bacteria 52,53 .Epetraborole demonstrated activity against wild-type A. baumannii (MIC 50 of 2 μg/mL) while it was less active against multidrug-resistant A. baumannii (MIC 50 of 8 μg/mL) 54 .In our study, epetraborole showed good activity against both isolates (MIC < 8 μg/mL) but more potent against A. baumannii 5075 (MIC = 5 μM or 1.37 μg/mL) compared to a greater resistant A. baumannii QS17-1084 (MIC = 20 μM or 5.47 μg/mL).In addition, epetraborole showed potent activity against Mycobacterium abscessus and the combination of epetraborole and norvaline improved the in vivo efficacy compared to epetraborole alone 54,55 .Brilacidin, a small-molecule arylamide mimics of antimicrobial peptides (AMPs), has exhibited potent bactericidal activity against drug-resistant and -susceptible strains of multiple Gram-negative and Gram-positive pathogens 56,57 .Brilacidin caused membrane depolarization in Staphylococcus aureus 58 and also used to be evaluated as an ocular anti-infective 59 .In addition, brilacidin showed the inhibitory activity against SARS-CoV-2 [60][61][62] and human pathogenic fungi 63 .Recently, MMV1580854, MUT056399, eravacycline, epetraborole, and MMV1579788 exhibited a higher percentage (> 90%) of Klebsiella pneumoniae growth inhibition 64 .This suggested the shared targets of the compounds between A. baumannii and K. pneumoniae.With respect to our IC 50 s of the compounds against both A. baumannii strains (Fig. 3), it was noticeable that some compounds were particularly potent against one strain.Doxycycline was less potent against A. baumannii QS17-1084.This may be due to the fact that A. baumannii QS17-1084 harbors tetracycline and macrolide resistant gene.Both strains harbor different beta-lactam resistant genes, and this might be accounted for the difference in the IC 50 values against imipenem.In terms of MMV1580854, brilacidin, and MUT056399, there is no confirmed resistant genes against these compounds.However, our results showed that A. baumannii QS17-1084 showed higher IC 50 values against both MMV1580854 and MUT056399, which share the same chemical scaffold.It might be worthwhile to check if there are mutations in fab1 gene of both strains since it was hypothesized that these two compounds should share the same target which is FabI.
In terms of drug combination, the combinations of among MMV1580854, MUT056300, brilacidin and eravacycline provided toward synergistic effects against our tested A. baumannii isolates.The synergism when www.nature.com/scientificreports/brilacidin combined with either MMV1580854 or MUT056399 could promote disrupting bacterial cell membranes and inhibiting fatty acid synthesis whereas when the combination of eravacycline combined with either MMV1580854 or MUT056399 could target for inhibiting both protein and fatty acid synthesis.The combination of eravacycline and brilacidin might disrupt protein synthesis and bacterial cell membranes.Brilacidin showed synergistic antiviral activity when combined with remdesivir 60 .Checkboard testing of one of the FabI inhibitors, sharing the similar structures as MMV1580854 and MUT056399 with known antibacterial agents against A. baumannii isolates revealed that the best effect was observed when the FabI inhibitor combined with colistin 65 The combination effects of eravacycline with other standard-of-care antibiotics were carried out against XDR A. baumannii and the combination between eravacycline and amikacin showed additive and synergistic 50 .The time-kill assay clearly confirmed the great efficacy of drug combinations in comparison to individual drugs.Bactericidal activities were observed significantly when drug combinations were used.The similar synergistic killing effect was also seen when econazole and colistin combination was treated to multidrug-resistant A. baumannii 66 .

Conclusion
We have identified 7 drug candidates that significantly suppressed the growth of the extensively drug-resistant Acinetobacter baumannii 5075 and QS17-1084 strains.We also found three pairs and two pairs of two drug combinations that exhibited toward synergistic effect against Acinetobacter baumannii 5075 and QS17-1084, respectively.These combinations arose from brilacidin, eravacycline and two triclosan derivatives, including MMV1580854 and MUT056399.Regards to time-kill and biofilm assays, the combinations of MUT056399 and brilacidin or eravacycline will be selected for the efficacy test in the chronic wound infection mice model.

AFig. 1 .
Fig. 1.Characteristics of A. baumannii ATCC19606 (ATCC19606), A. baumannii 5075 (AB5075), and A. baumannii QS17-1084 (QS17-1084).Strain sequence type (ST) is indicated next to each strain.The assigned antimicrobial agent resistance phenotype is provided where the green, yellow and blue squares indicate a result of susceptible, intermediate, and resistant to the tested antimicrobial agents, respectively.The known antimicrobial agent resistant genes are provided in which each color represents each class or subclass of antimicrobial agent as well as mode of their inhibition.The figure was produced using the iTOL tool 31 .

Fig. 2 .
Fig. 2. Antibacterial activities against A. baumannii QS17-1084 in the Pandemic Response Box collection at 10 µM concentration.Activities of known antibacterials are represented by blue circles [colistin (12.6 µM), erythromycin (12.8 µM), doxycycline (8.3 µM), imipenem (12.6 µM), and ciprofloxacin (9.4 µM)].Most of the compounds had activities below the 50% growth inhibition cutoff.Seven of the most active compounds were selected for further tests based on their ability to inhibit the growth of A. bamannii QS17-1084 in the 80-100% range.These compounds are indicated by red circles.

Table 1 .
Information on the compounds selected from the MMV Pandemic Response Box collection a .a MMV, Medicine for Malaria Venture.b Structures were redrawn using ChemDraw Professional.

Fig. 3 .
Fig. 3. Inhibition of A. baumannii in vitro growth by 7 MMV Pandemic response box compounds.(A) Half maximal inhibitory concentration (IC 50 ) of the compounds against A. baumannii QS17-1084 and 5075.(B) IC 50 comparison of MMV1580854 and MUT056399 against A. baumannii QS17-1084 and 5075.Statistically significant differences calculated by multiple unpaired t test are indicated with one (P < 0.05) and nd for no difference.

Table 2 .
Inhibitory and bactericidal properties of the selected MMV Pandemic response box compounds and control drugs.a MIC, minimum inhibitory concentration; MBC, minimum bactericidal concentration.