Skip to content
antibacterial .ai

Alternative approaches

Antimicrobial peptides: fast killing, hard development

AMPs kill by physically disrupting the membrane, which is why resistance is slow to appear and why selectivity is the whole game.

At a glance

Mechanism
Membrane disruption, often cationic and amphipathic
Clinical examples
Polymyxins including colistin, daptomycin
Advantage
Resistance is slower to appear
Liabilities
Haemolysis, protease stability, cost of goods

Antimicrobial peptides kill by physically disrupting the bacterial membrane rather than by inhibiting a single enzyme. That mechanism explains both why resistance is slow to appear and why development is unusually hard.

The mechanism

Most AMPs are short, cationic and amphipathic: a positive net charge attracted to the anionic bacterial surface, and a face of hydrophobic residues that inserts into the lipid bilayer. Bacterial membranes are rich in anionic phospholipids and, in Gram-negatives, wrapped in lipopolysaccharide, while mammalian outer leaflets are largely zwitterionic and contain cholesterol. That difference is the entire basis of selectivity.

Once bound, the accepted models describe barrel stave pores, toroidal pores or a carpet mechanism in which the peptide accumulates until the membrane loses integrity. Several AMPs also have intracellular targets once they get in, but membrane disruption is the primary event.

Why resistance is slower

Escaping a peptide that attacks the membrane means changing the membrane, which is expensive. Bacteria do it: modifying lipid A with phosphoethanolamine or aminoarabinose to reduce net negative charge, altering phospholipid composition, or producing proteases. But there is no single point mutation that abolishes binding the way a target mutation abolishes an enzyme inhibitor, so the frequency of resistance is generally low.

The exception proves the rule. Plasmid borne mcr-1 encodes exactly such a lipid A modification against colistin, and because it is plasmid borne it moves between organisms. Slow to appear is not the same as never.

The ones already in clinical use

  • Polymyxins, including colistin (polymyxin E). Cyclic lipopeptides that bind lipid A and permeabilise the outer membrane. Reserved because of nephrotoxicity, and still important against carbapenem resistant Gram-negatives.
  • Daptomycin, a calcium dependent lipopeptide acting on Gram-positive membranes. Inactive in the lung because pulmonary surfactant sequesters it, which is a good example of how membrane activity creates unusual failure modes.
  • Gramicidin and bacitracin, topical only, largely for toxicity reasons.

The liabilities, stated honestly

LiabilityWhy it happensWhat programs do about it
Haemolysis and cytotoxicityThe same amphipathicity acts on mammalian membranesOptimise the therapeutic index, measure it early and often
Protease degradationPeptides are substrates for host and bacterial proteasesD-amino acids, cyclisation, peptidomimetics
Serum and salt sensitivityDivalent cations and serum proteins compete with bindingAssay in physiological conditions, not just in low salt buffer
Poor oral exposureSize and chargeTopical, inhaled or intravenous routes
Cost of goodsPeptide synthesis at scaleShorter sequences, fermentation, mimetics

A common pattern in AMP papers is a peptide with an impressive MIC in low salt buffer and much weaker activity in physiological conditions. When reading published activity for an AMP series, the assay conditions matter as much as the number, which is a general point covered under minimum inhibitory concentration.

Screening an AMP series

Two things are worth knowing before you run one:

  1. Confidence will often be lower. The published neighbourhood for a novel peptide sequence is thinner than for a small molecule scaffold, and the honest output is a low confidence row rather than a confident number.
  2. The mechanism reasoning is still useful. Charge, amphipathicity and the presence of lipid A modifications in the target organism drive the call, and those are exactly the features a per strain read-out can talk about.

The screen below is preselected with colistin as a reference point. Replace it with your own sequence or structure and compare. For the alternative non chemical route to the same organisms, see phage therapy.

Run it

Screen a compound against the ESKAPE panel

The panel below is preselected for this page. Change the compound, run it, and read the matrix. Three screens per session, no account needed.

Run a screen

of 3 left

Strain panel

No account needed. 3 screens per session.

Ready to run Colistin ESKAPE panel
Nothing predicted until you run
Strain MIC (µg/mL) Call Resistance risk Conf.
Tick at least one strain in the panel on the left.

Nothing here is predicted yet

These are the strains this page preselected, waiting on a compound. Run the screen and every row fills in with an MIC band, an S / I / R call, the mechanism expected to decide it and a confidence out of four.

Strain MIC (µg/mL) Call Resistance risk Conf.

Why

Closest published analogs

Save this screen and run the rest of your series. screen left in this session. This session is used up.

Computational prediction from published literature. Research use only, not a lab measurement and not clinical guidance.

Take it from reading to a read-out

Run a screen on the compound this page made you think about, then create an account when you want to keep the result and bring the series.

Run a screen first

What you are agreeing to

  • Your compounds stay yours
  • Never used for model training
  • Deletable on request
  • No card required to run a screen

Screen your own compound.
No card required.