The Antibiotic Resistance Benchmark PAβN Gets a Long-Overdue Toxicity and Structural Makeover

For more than two decades, one small molecule has quietly underpinned nearly every effort to sabotage the molecular garbage disposal systems of Gram-negative bacteria. Phenylalanine-arginine beta-naphthylamide, known universally as PAβN, was discovered in a screen of a Japanese pharmaceutical collection for compounds that could restore the activity of levofloxacin against Pseudomonas aeruginosa. It works by jamming RND efflux pumps, the tripartite transporters that use the proton motive force to expel antibiotics before they can reach their intracellular targets. Although PAβN has long been described as too toxic for clinical use, it remains the admitted reference compound against which every new efflux pump inhibitor is judged. Now a team at Université Paris Cité and CNRS, working with colleagues in Besançon, has subjected this benchmark to a rigorous multi-approach audit, and the results are rewriting the rulebook on how it should be used.

Writing in Frontiers in Pharmacology, Hilal Wehbi, Elodie Olivier, Isabelle Broutin and their colleagues report that PAβN is far less toxic to human cells than its reputation suggests, at least under the conditions of classical laboratory exposure. The finding matters because PAβN is not merely a curiosity. It is the control molecule in virtually every phenotypic evaluation of RND efflux pump inhibitors, and any imprecision in its use propagates through the entire drug development pipeline. The team therefore set out to define, with unprecedented precision, the concentration limits at which PAβN can be deployed as a microbiological adjuvant, the toxicity thresholds across four human cell models, and the atomic-resolution details of how the molecule actually interacts with its transporter target.

The microbiological work centred on three Pseudomonas aeruginosa PA14 strains: the wild type, a mutant lacking the mexAB genes that encode the constitutive MexAB-OprM pump, and a nalC mutant that overproduces the same pump. Using broth microdilution following EUCAST recommendations, the researchers measured minimum inhibitory concentrations of three clinically important β-lactams, aztreonam, ceftazidime and ticarcillin, in the presence and absence of PAβN. The overproducer strain showed elevated MICs for all three antibiotics, consistent with enhanced efflux, while the deletion mutant showed reduced MICs. Adding PAβN at 50 and 100 micrograms per millilitre lowered the MICs in the wild type and overproducer strains, with the strongest effects at 100 micrograms per millilitre, where aztreonam and ceftazidime MICs fell fourfold and ticarcillin fell eightfold in the overproducer. Crucially, no MIC reduction was observed in the deletion mutant, confirming that the effect is genuinely MexAB-OprM-associated rather than a nonspecific antibacterial action.

The envelope perturbation experiments drew an equally sharp line. PAβN alone caused no detectable growth perturbation at 100 micrograms per millilitre, but growth effects appeared at 200 micrograms per millilitre and marked inhibition between 400 and 800 micrograms per millilitre. Combined with the absence of MIC reduction in the pump-deleted strain, this indicates that the frontier between PAβN’s two faces, competitive substrate inhibition and membrane permeabilization, lies between 100 and 200 micrograms per millilitre. The authors recommend that future studies use two or three concentrations within the safe range and verify envelope integrity with RND-deleted strains, since Pseudomonas aeruginosa is less permeable than Escherichia coli and the threshold may differ across Gram-negative species.

The toxicology results were the most surprising. The team exposed four human cell lines, bronchial epithelial BEAS-2B, alveolar carcinoma A549, embryonic kidney HEK293 and hepatocellular carcinoma HepG2, to PAβN across a concentration range from 50 to 1000 micrograms per millilitre, using seven complementary endpoints: metabolic activity by Alamar Blue, membrane damage by lactate dehydrogenase release, cell counting, propidium iodide uptake, reactive oxygen species production, mitochondrial membrane potential by JC-1, intracellular ATP levels and caspase activities. Across all four models, PAβN displayed limited cytotoxicity, with most effects confined to the highest concentrations approaching or exceeding the millimolar range. In contrast, β-naphthylamine, the carcinogenic fragment that would be released if PAβN’s amide bond were hydrolysed, was markedly more toxic, causing rapid oxidative stress, mitochondrial depolarization, ATP depletion and membrane injury at substantially lower concentrations.

This differential toxicity points to the β-naphthylamide moiety as a potential contributor to PAβN toxicity, possibly resulting from metabolic proteolysis processing in vivo. The authors stress that PAβN stability and β-naphthylamine formation were not directly assessed under their experimental conditions, and that LC-MS analysis would be required to confirm the hypothesis. Nevertheless, the pattern is striking: PAβN cannot be considered biologically inert toward mammalian cells, but its effects are strongly cell-type and endpoint dependent, and the dissociation between metabolic activity and membrane integrity that they observed is consistent with the heterogeneous effects previously reported in mammalian models. The practical implication is that PAβN’s clinical toxicity, if it exists, may stem not from the intact molecule but from fragments generated by physiological processing.

The structural biology component of the study is where the work becomes truly revelatory. Only one structure of a RND transporter in complex with PAβN had ever been solved, the AcrB structure from E. coli published in 2005 at 3.64 angstrom resolution. But that structure was obtained in a crystallographic space group H32 in which the trimer is reconstituted by a crystallographic threefold axis, forcing the trimer to be artificially symmetric and preventing the conformational changes required for substrate efflux. PAβN was found close to the entry of the efflux pathway rather than in the deep binding cavity, raising questions about whether the observed position reflects genuine biology or crystallographic constraint. The Paris team therefore solved new structures of MexB, the transporter of the constitutively expressed MexAB-OprM pump from Pseudomonas aeruginosa, in complex with PAβN by both crystallography and cryo-electron microscopy.

The crystallographic structure, obtained by soaking native MexB crystals with PAβN at 1 to 2 millimolar, revealed the inhibitor in the Tight monomer at an intermediate position between the proximal and distal binding pockets, just before the switch-Loop. But a molecule of the detergent n-dodecyl-β-D-maltoside, used for solubilization, occupied the deep binding pocket where structurally characterized inhibitors are usually found, apparently blocking PAβN from entering. Repeated co-crystallization attempts always yielded the same DDM molecule in the deep pocket, suggesting that DDM is itself a good MexB substrate present from the earliest purification steps. To circumvent this problem, the team reconstituted MexB in lipidic nanodiscs with POPC and the membrane scaffold protein MSP1E3D1, then solved the complex by cryo-EM at 2.31 angstrom resolution using a Titan Krios G4 microscope at the POLARIS facility at SOLEIL, processing 10483 movies down to 592436 particles.

In the cryo-EM structure, freed from detergent, PAβN was found in the deep binding pocket of the Tight monomer, stacked between Phe615 from the switch-Loop and F178, with the phenylalanine group surrounded by aromatic residues and the arginine group in a slightly less hydrophobic environment. The density was imperfect but clearly showed three branches corresponding to the three fragments of PAβN, indicating that the molecule is mobile within the cavity. Superposition of the two new MexB structures with the old AcrB structure revealed three very different positions of PAβN along the efflux pathway: at the entry in AcrB, in an intermediate resting zone in the crystal structure, and in the deep pocket in the cryo-EM structure. The conformation of the molecule differed radically in each case, with the three side chains not making contact in AcrB, the naphthylamide stacked on the phenyl in the MexB crystal structure, and the phenyl parallel to the arginine guanidinium in the cryo-EM structure.

These findings suggest that PAβN is not effluxed in a rigid conformation and that there is no real binding of the molecule with the protein before it reaches the deep binding pocket, but rather a series of resting zones along the path, consistent with its role as a competitive substrate rather than a true inhibitor. The comparison also validates the interacting region of PAβN by showing that the pyridopyrimidine inhibitor D13-9001 occupies the same deep pocket position. For drug designers, the message is clear: the cavities of the Tight monomer are highly adaptive, thanks to the loops E673-T678 and F666-P668 that reshape themselves to accommodate ligands, and future inhibitors must account for these resting zones rather than targeting a single static binding site. The study refines the use of PAβN as a reference efflux pump inhibitor, provides new perspectives on substrate transport by RND pumps, and underscores the importance of studying the stability of efflux pump inhibitors under proteolysis conditions to verify that resulting fragments will not be responsible for toxicity in therapeutic use.

Subject of Research: Revisit of the admitted control RND efflux pumps’ inhibitor PAβN: use limitations, and structural revelation in complex with MexB

Article Title: Revisit of the admitted control RND efflux pumps’ inhibitor PAβN: use limitations, and structural revelation in complex with MexB

Article References: Wehbi, H., Olivier, E., Cece, Q., Mégret, M., Sébastien, M., Aissaoui, N., Llanes, C., Phan, G., & Broutin, I. (2026). Revisit of the admitted control RND efflux pumps’ inhibitor PAβN: use limitations, and structural revelation in complex with MexB. Frontiers in Pharmacology, 17, Article 1906963. https://doi.org/10.3389/fphar.2026.1906963

Image Credits: AI Generated

DOI: 10.3389/fphar.2026.1906963

Keywords: Revisit, admitted, control, efflux, pumps, inhibitor, limitations, structural, revelation, complex, MexB, scientific research