The bacterial strains A. baumannii (ATCC BAA-1605, ATCC BAA-1710, and ATCC 17978), K. pneumoniae (ATCC BAA-1705), B. cenocepacia (ATCC-25416), and P. aeruginosa (ATCC BAA-2108) were purchased from the American Type Culture Collection (Manassas, VA, USA). Bacterial and fungal clinical isolates K. pneumoniae (VT-2646), K. oxytoca (VT-8943), B. cenocepacia (VT-2613), P. aeruginosa (VT-7530, VT-2824), P. fluorescens (VT-2535), Enterobacter cloacae (VT-1667, VT-4918), S. marcescens (VT-3904), Aeromonas hydrophila (VT-1409), Flavobacterium columnare (VT-5506), Ralstonia solanacearum (VT-5012), Xanthomonas campestris (VT-4657), Erwinia amylovora, (VT-3286), and Agrobacterium tumefaciens (VT-2679) were obtained from a private collection provided by Human Microbiology Institute (NY, USA).

{/it}All specimens were microbial suspensions derived from isolated colonies cultivated on agar plates using Columbia Blood Agar Base (Oxoid, ThermoFisher, USA). The suspensions were incubated in Mueller–Hinton broth (MHB) (Oxoid, ThermoFisher, USA) to guarantee uniform growth conditions for subsequent experimental analysis.

The MICs were determined using the broth macrodilution method, as described by the Clinical and Laboratory Standards Institute M27-A3 (CLSI) (Clinical and Laboratory Standards Institute, 2019). The MIC was defined as the lowest concentration of the antimicrobial agent that inhibits bacterial growth and possesses an optical density (OD570) increase ≤ 0.05 under cultivation conditions (Dong et al., 2012). Bacteria were cultivated in Mueller-Hinton broth at 37°C, except for Flavobacterium columnare, which was cultivated at 25°C for 24 h.

Resistance or sensitivity to antibiotics was evaluated based on the breakpoints established by the CLSI and European Committee on Antimicrobial Susceptibility Testing (EUCAST) guidelines, and it was dependent on the antibiotic used and the pathogen tested (CLSI, 2019, 2022).

The following antimicrobial agents were used in the study: Ampicillin, Amoxicillin, Piperacillin, Ceftriaxone, Ceftazidime, Cefepime, Aztreonam, Imipenem, Meropenem, Fosfomycin, Gentamicin, Tobramycin, Amikacin, Ciprofloxacin, Levofloxacin, Polymyxin B, Colistin, Doxycycline, Tetracycline, Chloramphenicol, Trimethoprim, Sulfamethoxazole, and Rifaximin (all from Sigma-Aldrich).

TGV-49 (Human Microbiology Institute) is a derivative of Mul-1867 with more polymeric chains, resulting in a higher level of antimicrobial activity ( Figure 1 ). For each study, the control samples were cultivated under the same conditions in antibiotic-free medium.

For the experimental evolution under TGV-49 pressure, we used the morbidostat device, custom-engineered based on general principles introduced by Toprak et al. (2013). Briefly, the morbidostat is a computer-controlled chemostat-like continuous culturing bioreactor where density of bacterial culture, which is constantly monitored by OD600, is controlled by varying an antibiotic concentration in media. Automated dilutions switch between addition of; (i) drug-containing media (increase of a drug concentration) when bacterial cultures are growing faster than dilution rate, and (ii) drug-free media (decrease of a drug concentration) when the growth rate slows down due to excessive drug pressure. This algorithm enables a gradually growing selective pressure driving the evolution of higher drug resistance. The detailed description of morbidostat implementation is provided on GitHub (https://github.com/sleyn/morbidostat_construction). Morbidostat-based experimental evolution of resistance to TGV-49 for A. baumannii ATCC 17978 was performed as previously described (Zlamal et al., 2021). Briefly, a morbidostat run was performed for 5 days upon inoculation of all six glass reactors with six individually prepared log-phase cultures derived from glycerol stocks of six individual colonies of A. baumannii ATCC17978 (Zlamal et al., 2021) at starting OD600 = 0.02 (20 mL in MHB/DMSO). During the run, the TGV-49 concentration in the drug-containing feed bottle was 200 mg/L (approximately 32xMIC of the unevolved parental strain). Glass reactor tubes were replaced every ~24 h (10 mL samples were taken) to minimize the impact of a biofilm accumulating on the walls of each reactor. For each reactor, glycerol stocks were prepared for all samples (taken daily) (B–F), and individual clones were randomly selected upon plating an aliquot of sample F on MHB-agar (15 clones from each reactor).

The obtained 90 clones were profiled for acquired resistance by growing them in 96-well microtiter plates in liquid MHB media without drug and supplemented with TGV-49 at 20 and 40 mg/L (corresponding to 1x and 2x MIC of unevolved strain, respectively). Clones with confirmed acquired resistance (≥ 2xMIC) were selected for the Illumina-based whole genome sequencing (WGS).

Genomic DNA was extracted using the GenElute bacterial genomic DNA kit (Sigma-Aldrich) using the manufacturer’s NA2110 protocol for Gram-negative bacteria. Libraries for sequencing were prepared using a NEBNext Ultra II FS DNA library prep kit for Illumina modules E7810L and E7595L (New England BioLabs) as per the manufacturer’s protocol (DNA input  ≥100 ng); TruSeq DNA PCR-free CD index 20015949 (Illumina) adapters were used to eliminate PCR amplification. Size selection and cleanup were performed using magnetic beads AMPure XP (Beckman Coulter). Prepared libraries were quantified using a NEBNext Library Quant kit for Illumina E7630L (New England BioLabs) and pooled with volumes adjusted to normalize concentrations for ∼200-fold genomic coverage. Pooled library size and quality were analyzed with the 2100 Bioanalyzer instrument (Agilent). Pooled DNA libraries were sequenced by Novogene Co. using HiSeq 4000 (A. baumannii clones) or HiSeq X10 (all other samples) instruments (Illumina) with paired-end 150-bp read length. Our computational pipeline used for WGS data analysis and variant calling has been previously described (Zlamal et al., 2021).

Data were analyzed using GraphPad Prism version 9.3.1. Two-way analysis of variance (ANOVA) was used for multiple comparisons at the 95% confidence level (p < 0.05). Data are presented as the mean ± standard deviation (SD) with three independent replicates. All experiments were performed in triplicate in three independent experiments.

The antimicrobial susceptibility testing demonstrated the potent broad-spectrum activity of TGV-49 against various Gram-negative bacterial strains. Notably, all tested multidrug-resistant clinical isolates (Acinetobacter baumannii, Klebsiella pneumoniae, Burkholderia cepacia, Pseudomonas aeruginosa, and Enterobacter cloacae) were resistant to a wide range of antibiotics, including beta-lactams, aminoglycosides, carbapenems, macrolides, and quinolones ( Figure 2A ). Some of the tested strains exhibited resistance even to last-resort treatments such as colistin. However, all these strains, including carbapenem-resistant Gram-negative bacilli from ESKAPE pathogens, were highly susceptible to TGV-49, with minimum inhibitory concentrations (MIC) ranging from 0.06 to 4.0 μg/mL.

Similarly, the antimicrobial testing of plant and fish pathogens (P. aeruginosa, P. fluorescens, A. hydrophila, A. tumerafaciens, F. columnare, R. solanacearum, X. campestris, and E. amylovora) demonstrated significant resistance to various common antibiotics used in humans ( Figure 2B ). However, these pathogens were susceptible to TGV-49 with low MIC values, comparable with those for human pathogens. The list of antibiotics used in this study and the results of susceptibility testing are presented in Figures 2A, B .

To investigate the dynamics of TGV-49 resistance development, we employed a morbidostat-based comparative resistomics workflow, as described by Zlamal et al. (2021), for clinically relevant Gram-negative bacteria, Acinetobacter baumannii ATCC 17978. Experimental evolution was performed in 6 parallel, 20 mL glass reactors of the custom-designed continuous culturing device (morbidostat), as described by Leyn et al. (2021). The procedure was performed under increasing pressure of TGV-49 drug at a concentration range of 0–200 mg/L as a function of growth rate assessed by constant monitoring of optical density (OD600). Five days of experimental evolution were sufficient for A. baumannii to acquire robust resistance against all tested clinical and experimental antibiotics (Zlamal et al., 2021; Leyn et al., 2023; Zampaloni et al., 2024). During this period, we observed no appreciable trend toward resistance against TGV-49. This observation was confirmed by MIC profiling of randomly picked clones (15 clones from each of the 6 reactors). Indeed, a mild resistance (2-4xMIC) was detected only among the clones from one (R5) of the six reactors.

We selected five clones (5F1–5F5, isolated from sample F taken from Reactor 5 close to the end of the 5-day evolutionary run) for analysis performed using Illumina-based whole-genomic sequencing (WGS). Data processing was performed by our computational pipeline (Zlamal et al., 2021), yielding and annotating Single Nucleotide Variants (SNV), short indels, Copy Number Variants (CNV), and insertion of IS-elements using the iJump algorithm (Zlamal et al., 2021). The results are listed in Supplementary Table S1A-C , and the significant variants are summarized in Supplementary Table S1D . All clones harbored essentially the same set of variants, including (i) SNV in the tRNA-Arg-CCG, (ii) IS-insert upstream of the pgsA gene, which encodes phospholipid biosynthesis enzyme (CDP-diacylglycerol–glycerol-3-phosphate 3-phosphatidyltransferase), and (iii) ~5–10-fold amplification of two genomic loci (24.3 and 13.5 Kb). Of these, the most likely hypothetical candidate for driver event is the amplification of the second locus, which contains three paralogous genes adeT1, adeT2, and adeT3 encoding a putative RND-type efflux pump; this has previously been implicated in aminoglycoside resistance (Magnet et al., 2001).

While the identity of the actual drug target(s) and resistance driver gene(s) (and, thus, the mechanisms of action and resistance) remain elusive, with the hypothesized impact of AdeT-driven efflux being a subject of prospective genetic exploration, this study points to a very low propensity of A. baumannii to acquire TGV-49 resistance.