Next-generation antibiotics against AMR

An antibiotic has to do two separate things: reach a sufficient concentration inside the cell, and bind a target once there. In Gram-positive organisms the second part is the hard one. In Gram-negatives it is almost always the first. That asymmetry is why the resistance crisis is not evenly distributed: it concentrates in Enterobacterales, Pseudomonas aeruginosa and Acinetobacter baumannii, the organisms with an outer membrane.

Enzymatic resistance: four classes, two chemistries#

Beta-lactams acylate the transpeptidase site of penicillin-binding proteins and so stop peptidoglycan cross-linking. Beta-lactamases hydrolyse the ring before it gets there. Under the Ambler scheme, classes A, C and D are serine hydrolases; class B are metalloenzymes carrying one or two zinc ions in the active site.

The distinction is not taxonomic housekeeping — it decides what can be inhibited at all. Diazabicyclooctanes (avibactam, relebactam) and boronates (vaborbactam) form a reversible covalent adduct with the catalytic serine, which is why they cover A, C and part of D and are useless against the metallo-beta-lactamases NDM, VIM and IMP: there is no serine to occupy. Inhibiting class B requires different chemistry aimed at the zinc itself, which is the design basis of broader-spectrum boronates such as taniborbactam. The practical consequence is that “carbapenem-resistant” is not one phenotype but at least two that are not therapeutically interchangeable.

The barrier and the pump work together#

The outer membrane is an asymmetric bilayer with lipopolysaccharide on its outer leaflet; hydrophobic molecules essentially do not cross it. Polar small molecules enter through porins — the water-filled beta-barrels OmpF and OmpC in E. coli, OprD in P. aeruginosa — under tight limits on size and charge. Losing or mutating a porin is on its own enough to shift the MIC into clinically relevant territory.

Running in parallel are the RND-family efflux pumps (AcrAB–TolC, MexAB–OprM), which expel a compound from the periplasm across the outer membrane in a single pass. This matters because barrier and pump do not add, they multiply. Slow entry combined with fast export gives an intracellular concentration of effectively zero, so improving one parameter alone changes nothing: a compound must simultaneously pass the porin and be poorly recognised by the pump.

The empirical accumulation rules for Gram-negative entry set out by the Hergenrother group in 2017 describe a narrow region of chemical space — a primary amine, low globularity (flat and rigid), moderate lipophilicity. Most chemical series that are optimal against a target do not sit in that region, and this, rather than a shortage of targets, is what explains the empty Gram-negative pipeline.

Workarounds and what they cost#

Cefiderocol solves transport rather than target engagement: its catechol group chelates ferric iron and the cell imports the complex through its own TonB-dependent siderophore receptors — active uptake in place of diffusion. The price is a fresh resistance route, through mutations in those iron-acquisition systems.

The overall ceiling stays evolutionary. Resistance arises either by mutation — target, porin, pump regulator — or by horizontal transfer of a ready-made gene on a plasmid. The second route is faster and indifferent to how clever the molecule is. The useful lifetime of a new class is therefore set less by its chemistry than by the volume of selective pressure it is placed under.