Introduction

Acinetobacter baumannii, classified as an aerobic Gram-negative pathogen, is widely recognized as a prominent cause of hospital-acquired infections and significant wound-related ailments1. These infections manifest diversely, encompassing conditions such as urinary tract infections, meningitis, bacteremia, gastrointestinal issues, as well as skin and soft tissue infections, including hospital-acquired and ventilator-associated pneumonia (HAP, VAP)1,2. Over the past decade, there has been a rise in treatment failures among patients infected with A. baumannii, coinciding with the global spread of A. baumannii strains resistant to multiple drugs, affecting regions worldwide, including Europe, North America, South America, Asia, and Southeast Asia2,3.

Multidrug-resistant (MDR) and extensively drug-resistant (XDR) A. baumannii present a formidable challenge in clinical settings. Carbapenems have conventionally served as the first-line drugs against these strains; however, the widespread use of carbapenems has led to an alarming increase in carbapenem-resistant A. baumannii (CRAB). Consequently, polymyxin (or colistin) has emerged as a last-resort treatment option for MDR and XDR A. baumannii infections, despite associated systemic toxicities such as nephrotoxicity and neurotoxicity2,4. Unfortunately, reports of increasing colistin resistance worldwide pose a further threat to effective treatment2,5,6.

Recognizing the critical nature of the situation, the World Health Organization (WHO) has designated CRAB as a high-priority antibiotic-resistant pathogen for the research and development of new antimicrobial agents7. Moreover, the Centers for Disease Control and Prevention (CDC) classifies CRAB as an "urgent threat," emphasizing the need for immediate attention and action in the US National Action Plan for Combating Antibiotic Resistant Bacteria and Department of Defense (DoD) guidelines, focusing on One-health and community-acquired infections (CAIs).

In Thailand, the prevalence of CRAB has surged to 70–80% in 2020, prompting a pressing need for the discovery of new drug candidates to address the growing crisis and control the spread of resistant strains8. Recognizing the complexity and length of the drug development process, our study aligns with the overarching goal of evaluating the Medicines for Malaria Venture's (MMV) Pandemic Response Box—a compilation of 400 drug-like compounds—as a potential source for novel antimicrobial agents9. Notably, previous investigations into the Pandemic Response Box have identified promising drug candidates for various infectious diseases, underscoring its potential significance in combating emerging threats10,11,12,13.

In this context, our primary objective is to evaluate the Pandemic Response Box collection against two extensively drug-resistant A. baumannii isolates. A previous screen of the Pandemic Response Box on other A. baumannii isolates identified five hits: MMV396785, MMV1578568, MMV1578574, MMV1578564, and MMV147985014. Our screening confirms these findings, except for MMV1479850. Additionally, we identified three new hits: MMV1580854, MMV1578566, and MMV1634402. Furthermore, we investigate drug combinations of the identified compounds to discover potential synergistic effects, aiming to develop innovative and effective treatment strategies.

Materials and methods

Compounds

In the pursuit of antibacterial activity against extensively drug-resistant A. baumannii, the MMV Pandemic Response Box compound library (Medicines for Malaria Venture, MMV) was systematically screened. Comprising 400 structurally diverse compounds, the Pandemic Box includes 201 antibacterial, 153 antiviral, and 46 antifungal compounds15. These compounds, provided in 96-well plates at a concentration of 10 mM in dimethyl sulfoxide (DMSO), have either been marketed or are in various phases of drug discovery or development.

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, USA16,17. 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 AHAH0000000016. 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 guidelines19,20. 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:

$$\frac{OD600\; of \;compound \; treated \;A. \;baumannii-OD600 \; of \; the \;blank \;well) \times 100}{(OD600 \;of \;DMSO\; treated \;A. \;baumannii-OD600 \;of\; the \;blank\; well)}$$

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 software22,23.

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 Standards Institute (CLSI) guidelines21. 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 dilutions20. 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 × 105 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 IC50 (50% inhibitory concentration) determination, A. baumannii strains were treated with various concentrations of test compound and cultured according to the microdilution method above. IC50s 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 modifications24,25,26. 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 × 104 CFU per well were added, and the test plates were incubated for 20–24 h at 37 °C. Bacteria growth was measured by OD600 as described above and the individual 50% fractional inhibitory concentrations (FIC50) were determined as previously described27. Isobolograms were constructed by plotting the FIC50 of drug A against the FIC50 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 FIC50A and FIC50B. The sum FIC50 (ΣFIC50) values indicate the kinds of interaction as follows: synergism when ΣFIC50 < 0.5; toward synergism when ΣFIC50 < 1; additive when ΣFIC50 = 1; toward antagonism when ΣFIC50 > 1; antagonism when ΣFIC50 > 2 to 4. The IC50s 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 OD600 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 log10, but < 3 log10 was defined as bacteriostatic activities, and the reduction of ≥ 3 log10 was defined as bactericidal activities, relative to initial inoculum. For drug combination, the reduction of ≥ 2 log10 relative to the most active component alone was defined as synergistic activities28. 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 OD595. 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.

$$\text{Inhibition \;of \;biofilm \;formation }(\text{\%}) =1- \left[\frac{OD595 \;of \;compound\; treated\; A.\; baumannii}{OD595 \;of \;untreated \;A.\; baumannii} \right] \times 100$$

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.

Results

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 models29,30. 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.

Fig. 1
figure 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 tool31.

Full size image

Antimicrobial resistance genes detection

A. baumannii QS17-1084 was the further characterized through advance genome sequencing in comparison with A. baumannii ATCC19606 and 5075 (Fig. 1). A. baumannii QS17-1084 was classified in sequence type (ST) 2, while A. baumannii 5075 was ST1. Genes causing resistances to 9 antibiotic classes were identified. A.baumannii QS17-1084 carried genes causing resistance to aminoglycosides (ant(3″)-IIa, aph(3″)-Ib, and aph(6)-Id and armA), β-lactams (blaADC-73, blaOXA-23, blaOXA-66, and ftsI(A515V)), fluoroquinolones (gyrA(S81L) and parC(S84L)), macrolides (mphE and msrE), tetracyclines (tetB), and fosfomycin (abaF). In addition, adeC and amvA involving in efflux pump over activity and causing reducing susceptibility or resistance to several antimicrobial classes such as β-lactam aminoglycosides, fluoroquinolones, tetracyclines, and erythromycin, respectively, as well as nreB gene generating resistance to heavy metals were also detected in A. baumannii QS17-1084. Similar to A. baumannii QS17-1084, A. baumannii 5075 harbored genes encoding resistance to aminoglycosides (aac(6′)-Ib3, aadA2, ant(2″)-Ia, ant(3″)-IIa, aph(3″)-Ib, aph(3′)-VIa, and aph(6)-Id), β-lactams (blaADC-11, blaGES-11, blaOXA-23, and blaOXA-69), fluoroquinolones (gyrA(S81L) and parC (S84L)), sulfonamides (sul1), trimethoprim (dfrA7), phenicols (cmlA1), fosfomycin (abaF), heavy metal (nreB), and biocide (qacEΔ1). It is noted that A. baumannii 5075 was susceptible to tetracyclines.

Identification of screening drugs against extensively drug-resistant A. baumannii

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).

Fig. 2
figure 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.

Full size image
Table 1 Information on the compounds selected from the MMV Pandemic Response Box collectiona.
Full size table

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 may relate to alexidine, a symmetrical alkyl bisbiguanide whereas brilacidin, an investigational new drug representing a new class of antimicrobial agents called host defense protein mimetics, 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 the MIC values of 40 and 50 µM, respectively. The compounds expressing the MIC value greater than 100 µM included doxycycline (133 µM), MMV1580854 and MUT056366 (> 160 µM). Unlike A. baumannii QS17-1084, the MIC values of the four compounds considerably decreased when tested against A. baumannii 5075. The MIC value of doxycycline was 133-fold decrease, followed by eravacycline (16 folds), epetraborole (fourfolds), and alexidine (twofolds). The highest MIC values were still from MMV1580854 and MUT056399 (> 160 µM). Most MBC values were similar to the MIC values except for epetraborole of A. baumannii QS17-1084 and imipenem of A. baumannii 5075 in which their MBC values were higher than their MIC values (> 160 µM for epetraborole and 100 µM for imipenem).

Table 2 Inhibitory and bactericidal properties of the selected MMV Pandemic response box compounds and control drugs.
Full size table

To better understand drug sensitivity, we subsequently determined the IC50 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 IC50 values ranging from 1.61 to 1.99 µM. Four compounds including alexidine, MMV1580854, brilacidin, and imipenem provided a comparable IC50 values ranging from 4.22 to 7.17 µM, while MUT056399 and doxycycline had the elevated IC50 values of 12.14 and 84.87 µM, respectively. When the compounds were tested against A. baumannii 5075, the IC50 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 IC50 value, followed by MUT056399 (8.20-fold decrease), and MMV1580854 (2.71-fold decrease). Two compounds, brilacidin and imipenem, exhibited the increased IC50 values against A. baumannii 5075 around 1.94- and 1.52-fold increases, respectively. In addition to IC50 values, IC99 values of each compound were also determined (Table S1). The IC99 concentration which is approximately 100-fold the IC50 concentration is used when complete inhibition is required. The IC99 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 IC99 values were smaller than the MIC values. In the case of A. baumannii 5075, the IC99 values of eravacycline, epetraborole, and doxycycline were close to the MIC values. While the IC99 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 IC50 values (Fig. 3 B). When both compounds were tested against A. baumannii 5075, the IC50 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 IC50 value of MMV1580854 was 2.87-fold decreased compared to the IC50 value of MUT056399, indicating that MMV1580854 was more potent than MUT056399 when treated against A. baumannii QS17-1084.

Fig. 3
figure 3

Inhibition of A. baumannii in vitro growth by 7 MMV Pandemic response box compounds. (A) Half maximal inhibitory concentration (IC50) of the compounds against A. baumannii QS17-1084 and 5075. (B) IC50 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.

Full size image

Drug combination effects on A. baumannii

Finding novel antimicrobial agents and therapies based on synergistic combinations are crucial to battle drug-resistant 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 methods25. Two drug combinations (brilacidin-MMV1580854, and brilacidin-MUT056399) showed toward synergistic effects (ΣFIC50 < 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 (ΣFIC50 > 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.

Fig. 4
figure 4

Drug interactions in A. baumannii. Forty-six drug combinations were carried out against both A. baumannii QS17-1084 and 5075. The numbers represent ΣFIC50 (50% Fractional Inhibitory Concentrations) values: ΣFIC50, synergism when ΣFIC50 ≤ 0.5; toward synergism when ΣFIC50 < ; additive when ΣFIC50 = 1; toward antagonism when ΣFIC50 > 1; antagonism when ΣFIC50 ≥ 2 to 4. The values show the mean ± S.D. of 3 independent assays for each A. baumannii strain.

Full size image

Time-kill assay for assessing the efficacy of drug combinations

According to drug combination results, in vitro time-kill assay was performed to evaluate drug combination antibiotic efficacy. The drug combination with synergistic effect was chosen (Fig. 4). For A. baumannii QS17-1084, MUT056399 and brilacidin at the ratios of 1:2 (low FIC50) and 4:1 (high FIC50), respectively, were selected (Table S2), whereas MUT056399 and eravacycline were chosen at the ratios of 1:2 (high FIC50) and 3:1 (low FIC50) (Table S3) to test against A. baumannii 5075. Table S4 summarizes the concentration used in the assay. Figure 5 shows time-kill curves of A. baumannii strains against both individual drug and combined drugs. According to Fig. 5A, A. baumannii QS17-1084 treated with the combination of MUT056399 and brilacidin at the ratio 1:2 shows the complete inhibition after 24 h with the reduction of 5.5 log10 CFU/mL bacteria compared to the initial count (P value < 0.0001) and the reduction of 4 log10 CFU/mL compared to the most active compound alone, 10 µM brilacidin, (P value < 0.0001). The observed reduction indicated bactericidal and synergistic activities of MUT056399 combined with brilacidin at a ratio of 1:2. In contrast, the combination of MUT056399 and brilacidin at the ratio 4:1 did not profoundly inhibit the QS17-1084 strain and showed a reduction by 10.6% after 24 h (5.05 log10 CFU/ml) compared to initial inoculum (5.65 log10 CFU/ml) (Fig. 5A). In addition, CFU/mL of the drug combination MUT056399 with brilacidin at the ratio 4:1 was not significantly different compared to each drug alone. Furthermore, MUT056399 and brilacidin alone as well as their combination at the ratio 4:1, showed neither bacteriostatic nor bactericidal effects against A. baumannii QS17-1084.

Fig. 5
figure 5

Time-kill curve. (A) A. baumannii QS17-1084 incubated without antibiotic (growth control), with MUT056399, brilacidin, MUT056399:brilacidin (ratio 1:2), and MUT056399:brilacidin (ratio 4:1). (B) A. baumannii 5075 incubated without antibiotic (growth control), with MUT056399, eravacycline: MUT056399:eravacycline (ratio 3:1), and MUT056399 : eravacycline (ratio 1:2). *P < 0.05 was analyzed by Dunnett's multiple comparisons test to compare between drug combination (MUT056399 and brilacidin at the ratio 1:2 for A. baumannii QS17-1084; and MUT056399 and eravacycline at ratio 1:2 for A. baumannii 5075) versus other tested compounds and growth control.

Full size image

For A. baumannii 5075, the combination of MUT056399 and eravacycline exhibited a significant reduction in CFU/mL at the ratio of 1:2 with the reduction of 3.8 log10 CFU/mL) from the initial inoculum after 24 h (P value < 0.0001; Fig. 5B) and with the reduction of 2.2 log10 CFU/mL) in comparison to eravacycline alone (P value < 0.0001; Fig. 5B), shown as the most active compound. In addition, this drug combination at the ratio 1:2 displayed the synergistic and bactericidal activities. However, it was observed that combination of MUT056399 and eravacycline at the ratio of 3:1 did not present bactericidal activities. It seems that this minimal inhibitory effect was caused by eravacycline alone (Fig. 5B).

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).

Fig. 6
figure 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.

Full size image

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 pneumonia33. The 400 MMV Pandemic Response Box compounds were screened, and 7 compounds were selected and further evaluated through phenotypic experiments against 2 extensively 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 aminoglycosides36. 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 blaADC belongs to class C β-lactamases. The ADC β-lactamase are cephalosporinase with resistance extended-spectrum cephalosporins37. The blaoxa genes, classified for class D β-lactamases resistance, were reported in most of carbapenemase producing A. baumannii in several studies in Southeast Asia, including Thailand38, Vietnam39, Malaysia40, China41 as well as in Europe42. The blaoxa-66 was grouped into intrinsic blaoxa-51-like oxacillinase gene found in A. baumannii37. 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 IC50 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 apoptosis43. It also inhibited planktonic growth, prevented biofilm formation as well as killing biofilms formed by diverse fungal organisms44. Alexidine dihydrochoride was also identified in a quantitative high throughput screen (qHTS) against A. baumannii 5075 with the IC50 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 saprophyticus45. MMV1580854 and MUT056399, which are diphenyl ether derivatives, have been shown as inhibitors against enoyl-acyl carrier protein reductase (FabI)46,47. The IC50 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 MUT05639947. In our study, while the IC50 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 Acinetobacter48, inhibits bacterial protein synthesis through binding to the 30S ribosomal subunit. Eravacycline exhibited greater activity than the comparators of the tetracycline class with the MIC50 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 isolates50. 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 (IC50 ranging from 0.39 to 66.5 µM)51. Tigecycline, a derivative of the tetracycline, was the most potent with the IC50 of 0.15 µM. This was in accordance with our result that eravacycline was the most potent with the IC50 of 0.22 µM51. Epetraborole, a new class of leucyl-tRNA synthetase inhibitors, showed potent activity against a broad range of Gram-negative bacteria52,53. Epetraborole demonstrated activity against wild-type A. baumannii (MIC50 of 2 μg/mL) while it was less active against multidrug-resistant A. baumannii (MIC50 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 alone54,55. 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 pathogens56,57. Brilacidin caused membrane depolarization in Staphylococcus aureus58 and also used to be evaluated as an ocular anti-infective59. In addition, brilacidin showed the inhibitory activity against SARS-CoV-260,61,62 and human pathogenic fungi63. Recently, MMV1580854, MUT056399, eravacycline, epetraborole, and MMV1579788 exhibited a higher percentage (> 90%) of Klebsiella pneumoniae growth inhibition64. This suggested the shared targets of the compounds between A. baumannii and K. pneumoniae. With respect to our IC50s 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 IC50 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 IC50 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 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 remdesivir60. 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 colistin65 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. baumannii66.

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.