An amphipathic peptide with antibiotic activity against multidrug-resistant Gram-negative bacteria

Multidrug-resistant (MDR) Gram-negative bacteria pose a critical threat due to their impermeable outer membrane, efflux pumps, and accumulated resistance mechanisms against all major antibiotic classes. While membrane-targeting lipopeptides such as colistin are used as last-resort agents, their narrow therapeutic index and rising resistance (e.g., plasmid-mediated mcr-1) limit clinical utility. Antimicrobial peptides (AMPs) offer an alternative modality, but many are cationic and exhibit cytotoxicity. Rules for small-molecule penetration into Gram-negative bacteria (size, polarity, charge, rigidity) have guided discovery; analogous design principles for peptide antibiotics are less established. The authors aimed to apply structure-based optimization to an amphipathic β-hairpin AMP, arenicin-3, to enhance Gram-negative antibacterial activity and in vivo efficacy while reducing mammalian toxicity.

Background highlights include: (1) Only polymyxin lipopeptides (colistin/polymyxin B) are clinically approved membrane-targeting antibiotics for Gram-negative infections, but with nephrotoxicity and emerging resistance, including plasmid-mediated mcr-1. (2) Some AMPs (e.g., LL-37) exhibit immunomodulatory effects with limited direct in vitro potency but notable in vivo benefits. (3) Physicochemical rules for Gram-negative penetration by small molecules emphasize low molecular weight, polarity, positive charge, and rigidity; analogous peptide design principles suggest balancing amphipathicity, charge distribution, and hydrophobicity to optimize activity and minimize toxicity. (4) Arenicin-3 is a 21-residue amphipathic β-hairpin with two disulfide bonds, potent in vitro activity but cytotoxicity and hemolysis; related analogs (e.g., NZ17074) disrupt membranes and affect multiple bacterial pathways. This context motivates rational peptide redesign to achieve efficacy and safety against MDR/XDR Gram-negative pathogens.

- Structural determination and design: Determined the NMR solution structure of arenicin-3 (PDB 5V0Y; BMRB 30259), revealing a twisted β-hairpin stabilized by two disulfide bonds (Cys3–Cys20, Cys7–Cys16) forming an amphipathic surface. Conducted alanine scanning and residue substitutions/deletions to modulate lipophilicity and charge, generating analog panels. Designed AA139 (sequence: GFCWYVCARRNGARVCYRRCN; V8A, Y9R, V13A) based on structure-activity and structure-toxicity relationships. - In vitro susceptibility: Determined MICs by broth microdilution (MHB/CAMHB) across reference strains and clinical isolate panels (US panel n=331; worldwide panel n=445) including non-MDR, MDR, and XDR isolates. Assessed activity in presence of 50% human serum and 5% lung surfactant. Agarose MICs corroborated broth results. - Cytotoxicity/hemolysis: Measured CC50 in HEK-293, HepG2, HK-2 cells (Alamar Blue) and hemolysis (HC10/HC50) on human erythrocytes. Tested AA139 on primary human hepatocytes (IC50) after 24 h. - Membrane interaction assays: Surface plasmon resonance on DMPC and DMPC+10% E. coli lipid A bilayers to assess lipid binding. Permeabilization/depolarization assays included NPN uptake (outer membrane), DiSC3(5) (cytoplasmic membrane potential), and SYTOX Green uptake with flow cytometry (inner membrane permeability). Transmission electron microscopy visualized ultrastructural changes in E. coli and P. aeruginosa after peptide exposure. - ATP release: Quantified extracellular ATP over time after exposure to MIC-multiples of arenicin-3, AA139, colistin, and piperacillin (BacTiter-Glo). - Resistance propensity: Estimated spontaneous mutation frequency to AA139 versus colistin at 4× and 8× MIC in E. coli, K. pneumoniae, and P. aeruginosa. Performed 20-day serial passage resistance induction in two isolates each of E. coli, K. pneumoniae, A. baumannii, and P. aeruginosa, followed by 3 drug-free passages to assess stability. - Genetics: TraDIS on a dense Tn5 library of UPEC ST131 NCTC 13441 exposed to 0.25× MIC arenicin-3 to identify genes affecting fitness under selection; mapped insertions and analyzed with Bio-Tradis. Whole-genome sequencing of serially passaged E. coli ATCC 25922 isolates to detect resistance-associated mutations. - In vivo efficacy: Neutropenic mouse peritonitis and thigh infection models challenged with MDR E. coli AID#172; single or two-dose IV AA139 regimens compared to meropenem or polymyxin B. Mouse P. aeruginosa ATCC 27853 pneumonia model treated with aerosolized AA139 at varying concentrations and exposure durations (comparative to tobramycin). UTI model with ESBL E. coli DSA 443 treated with twice-daily IV AA139 versus meropenem. - Pharmacokinetics: IV 2 h infusion studies of AA139 in mice, cynomolgus monkeys, and minipigs; single IV bolus in mice. Quantified plasma/serum concentrations by LC-MS/MS; non-compartmental analysis. Inhalation PK in mice measured epithelial lining fluid, lung tissue, and plasma concentrations including day 1 and day 7 once-daily dosing. - Toxicology: Single and repeated IV dosing studies in mice, monkeys, and minipigs; 7-day inhalation toxicity in mice. Determined NOAELs and evaluated target organ toxicities histopathologically.

- Lead optimization: Structure-guided substitutions yielded AA139 (V8A, Y9R, V13A) with retained amphipathic β-hairpin and improved selectivity. - Potent Gram-negative activity: AA139 MICs (µg mL⁻1) against reference and resistant strains included E. coli ATCC 25922 = 0.125; K. pneumoniae BAA-2146 (NDM-1) = 1. In large clinical panels, AA139 MIC90 values were E. coli = 1.0, K. pneumoniae = 4.0, P. aeruginosa = 8.0, A. baumannii = 2.0, largely independent of resistance phenotype (non-MDR, MDR, XDR). - Activity vs colistin-resistant isolates: No cross-resistance with colistin. Among colistin-resistant isolates, AA139 MICs did not exceed 0.5 µg mL⁻1 (E. coli, colistin MIC 4–16), 2 µg mL⁻1 (A. baumannii, colistin MIC 8–232), 8 µg mL⁻1 (P. aeruginosa, colistin MIC >32), and 4 µg mL⁻1 for most K. pneumoniae (colistin MIC 8–232; two isolates higher). - Serum/surfactant effects: Typically 8–16-fold MIC increase in 50% human serum and 2–8-fold in 5% lung surfactant; AA139 retained useful activity in surfactant. - Reduced toxicity: AA139 showed no detectable cytotoxicity in HEK-293 and HepG2 (>300 µg mL⁻1) or HK-2 (>250 µg mL⁻1), and no hemolysis at >300 µg mL⁻1. Primary human hepatocyte IC50 = 130 µM (≈330 µg mL⁻1). This corresponds to 300–2400× the MIC against K. pneumoniae and E. coli, respectively. - Membrane interactions: SPR showed arenicin-3 variants bind DMPC independently of lipid A, unlike polymyxins. NPN assays demonstrated OM permeabilization; DiSC3(5) indicated weak depolarization; SYTOX Green uptake confirmed inner membrane permeabilization and cell death. TEM revealed membrane disruption and cytoplasmic leakage, more pronounced in P. aeruginosa. - Rapid ATP leakage: At 60 min and high multiples of MIC, arenicin-3 and AA139 induced ~93% and ~90% ATP release, respectively, exceeding colistin (~80%) and piperacillin (~20%). - Low resistance propensity: Spontaneous mutant frequency with AA139 was ≤1.6×10⁻9 (4× MIC) and ≤6.7×10⁻10 (8× MIC) across tested strains, markedly lower than colistin in E. coli and K. pneumoniae (up to ~3.9×10⁻7). Serial passage yielded modest, often non-stable MIC increases in E. coli (up to 8–16-fold), but higher and stable increases in K. pneumoniae (16–64-fold). - Genetic determinants: TraDIS under arenicin-3 selection enriched insertions in mlaA–F (log2 fold change ~4.95–5.3), indicating increased fitness of mla mutants; outer membrane integrity genes (slyB, yfgB) and efflux (macAB, tolC) showed decreased insertions. WGS of serially passaged E. coli identified a mlaC point mutation (T33G; L11R), supporting involvement of phospholipid transport in mechanism/adaptation. - In vivo efficacy: In neutropenic mouse peritonitis (MDR E. coli), single IV AA139 dose ED50 = 1.85 mg kg⁻1 (peritoneal fluid) and 1.55 mg kg⁻1 (blood); 3.75 mg kg⁻1 gave ~3-log (peritoneal fluid) and ~4-log (blood) CFU reductions vs vehicle, outperforming meropenem 40 mg kg⁻1. In the thigh model, AA139 ED50 = 4.8 mg kg⁻1; 10 mg kg⁻1 achieved 3.61-log reduction (comparable to polymyxin B 5 mg kg⁻1). In P. aeruginosa pneumonia, aerosolized AA139 30 mg mL⁻1 for 30 min reduced lung burden by 6.6 log CFU g⁻1, with 6/7 mice below detection. In UTI (ESBL E. coli), twice-daily IV AA139 5 mg kg⁻1 over 2 days significantly reduced bacterial loads in urine, bladder, and kidneys, comparable to meropenem 40 mg kg⁻1. - Pharmacokinetics: After IV administration, AA139 exhibited rapid distribution, bi-exponential elimination with t1/2 ~2–4 h, and moderate clearance suggestive of renal elimination. Inhaled AA139 achieved high epithelial lining fluid and lung tissue levels with minimal systemic exposure; concentrations persisted up to 24 h post-dose, supporting potential once-daily dosing. - Safety: Dose-limiting systemic toxicity consisted mainly of exaggerated pharmacological effects; kidney tubular nephropathy observed in minipigs at 20–30 mg kg⁻1 day⁻1 (2 h IV infusion for 7 days). NOAELs: mice 20 mg kg⁻1 day⁻1 IV (≈10× efficacious dose), monkeys 15 mg kg⁻1 day⁻1 IV, and mice 20 mg kg⁻1 day⁻1 via inhalation.

The study demonstrates that rational, structure-guided optimization of an amphipathic β-hairpin AMP can decouple antibacterial potency from mammalian cytotoxicity by tuning amphipathicity, hydrophobicity, and charge. AA139 retains strong Gram-negative activity, including against MDR/XDR and colistin-resistant isolates, while exhibiting markedly improved safety margins in vitro and in vivo. Mechanistic studies indicate primary interaction with bacterial membranes independent of lipid A, leading to outer membrane permeabilization and inner membrane compromise with rapid ATP leakage. Genetic analyses implicate the mla phospholipid transport system in bacterial adaptation, consistent with a mechanism tied to membrane homeostasis and outer membrane asymmetry. In vivo efficacy across systemic (peritonitis, thigh) and localized (pneumonia via inhalation, UTI) infection models, combined with favorable PK and defined NOAELs, supports the translational potential of AA139 as a peptide antibiotic targeting Gram-negative pathogens.

This work provides a structure-based pathway to develop bacteria-selective peptide antibiotics with broad Gram-negative activity, reduced cytotoxicity, and demonstrable in vivo efficacy. AA139 emerged as a lead candidate active against MDR/XDR and colistin-resistant E. coli, K. pneumoniae, A. baumannii, and P. aeruginosa, effective in murine models of peritonitis, pneumonia, and UTI, with manageable PK and safety profiles (NOAEL ≈10× efficacious doses in mice). Future research could focus on: (1) advanced preclinical safety and toxicology, including renal risk mitigation strategies; (2) elucidating molecular details of AA139 interactions with phospholipid transport (mla system) and membrane components; (3) optimization of inhalation delivery with defined deposited doses; (4) resistance surveillance and combination strategies; and (5) IND-enabling studies and early-phase clinical trials.

- In the aerosolized pneumonia model, the exact delivered pulmonary dose was not quantified, limiting dose-exposure interpretation. - While AA139 showed low spontaneous resistance frequency, serial passage induced stable MIC increases in K. pneumoniae (16–64-fold), indicating species-dependent resistance risk. - Activity decreased in human serum (typically 8–16-fold MIC increase), which may impact systemic free-drug exposure requirements. - Toxicology indicated kidney as a target organ at high systemic exposures (tubular nephropathy in minipigs), necessitating careful dose selection and monitoring. - The mechanism, while consistent with membrane perturbation and involvement of the mla system, remains to be fully elucidated at molecular resolution.