Chapter 25 — Key Takeaways

Antimicrobial Peptides: The Natural Weapons That Could Solve Antibiotic Resistance

The ten sentences

1. Roughly 1.27 million deaths were directly attributable to bacterial antimicrobial resistance in 2019; the associated figure of about 4.95 million counts something different, and a source that does not say which one it means does not know.

2. Antimicrobial peptides are made by essentially every multicellular organism — typically 12–50 residues, usually cationic, characteristically amphipathic — as part of innate immunity, which makes them a defense system hundreds of millions of years old.

3. They kill by disrupting membranes, not by binding a target. A positive charge is drawn to a negatively charged bacterial surface; a greasy face then inserts into the bilayer and permeabilizes it. There is no receptor, no amplification, and no lock-and-key specificity.

4. Resistance to membrane disruption is harder to evolve — not impossible. A point mutation can defeat a target-specific antibiotic; restructuring a membrane is a much larger evolutionary ask. But surface charge modification, efflux, proteolysis, and shielding are all documented, and mcr showed that colistin resistance can travel between organisms on a plasmid.

5. The three families to know: defensins (α in neutrophils and Paneth cells, β in epithelia), cathelicidins (in humans, only LL-37, which is also strongly immunomodulatory), and magainins (frog skin, 1987, the compounds that opened the field to drug development).

6. Selectivity is the field's central obstacle. Bacterial and mammalian membranes differ in charge, cholesterol, and transmembrane potential — but by degree, not by kind. Hemolysis is the standard toxicity readout, the selectivity index is the ratio that matters, and a great many excellent peptides have failed on that single axis.

7. The economics are as much of the barrier as the biology. Peptides are costly to manufacture, and antibiotics are commercially the worst class in medicine: used briefly, used rarely, and deliberately withheld by correct stewardship. A drug used briefly, rarely, and only when others fail cannot recover its development costs, and no scientific advance fixes that.

8. Topical AMP applications have gone much further than systemic ones, which follows directly from the selectivity argument — and as of this writing, no AMP has become a widely used systemic antibiotic.

9. Peptide antibiotics are already in clinical use and have been for decades. Colistin, daptomycin, vancomycin, polymyxin B, bacitracin, gramicidin. The category is not speculative; what has not yet succeeded is the narrower project of turning innate immunity's host defense peptides into new broad-spectrum systemic drugs.

10. 🔬 is not ❌. ❌ says a confident claim has outrun its evidence. 🔬 says the science is early, proceeding properly, and too soon to rate — and an honest 🔬 names, bidirectionally, what would resolve it.

The ratings from this chapter

All assessed as of this writing, 2026. Note the shape of the table: one category, five claims, four different ratings. That is the rating system working, not failing.

The two sentences to correct in other people's writing

Wrong: "Antimicrobial peptides don't work as drugs yet." Right: "Peptide antibiotics work, are approved, and are in daily use. The specific project of converting innate immunity's host defense peptides into new broad-spectrum systemic antibiotics has not yet produced an approved agent, for reasons that are partly biological and partly economic."

Wrong: "Bacteria can't develop resistance to antimicrobial peptides." Right: "Resistance to membrane-active peptides appears harder to evolve and costlier to maintain than resistance to target-specific antibiotics — which is a real advantage, and is not immunity."

The transferable skills

Ask which number. Attributable or associated, absolute or relative, incidence or prevalence. Any figure doing rhetorical work should be interrogated for what it counts before it is used.

Ask what the comparator was. Pexiganan failed in part because the comparator arm did well. A trial is a comparison, and choosing an easy comparator is one of the commonest ways a weak agent is made to look strong.

Ask about the margin, not the potency. Two compounds with identical MICs can have completely different futures, and the number that decides it is not in the antibacterial data.

Watch for a conspiracy replacing a technical problem. When someone explains a stalled field by naming a villain, go looking for the physical obstacle. It is usually still there, described in detail, in the field's own literature.

Never let a ✅ read as an endorsement. Colistin causes kidney injury and is rated ✅. A rating assesses evidence for a claim; it does not describe how pleasant a drug is to receive.

Where this goes next

• Chapter 18 returns to LL-37's immunomodulatory role, where the mechanism is strong and the clinical results have been considerably less impressive than the mechanism suggested.
• Chapter 30 applies the topical-versus-systemic selectivity argument to cosmetic peptides, where the same physics produces a different set of claims.
• Chapters 32 and 33 cover peptide manufacturing and engineering — the two levers most likely to move the 🔬 in this chapter, if it moves.
• Chapter 34 covers what is actually in an unregulated vial, which is a separate question from whether the peptide named on the label would be safe.