Defensins, cathelicidins and why antimicrobial peptides did not make it as antibiotics
The idea is old and biologically attractive: the immune system produces its own antimicrobial peptides — defensins, cathelicidins, magainins — that kill bacteria directly, often including resistant ones. With the growing global antibiotic resistance crisis they seem the obvious solution. Despite 40 years of research and hundreds of clinical trials, not a single systemically approved antimicrobial peptide drug exists. A stocktake of the structural reasons.
The biology: an ancient defence system
Antimicrobial peptides (AMPs) are an evolutionarily ancient defence system — in plants, insects, frogs, mammals. They are short (typically 10-50 amino acids), often positively charged, amphipathic (with hydrophobic and hydrophilic sides), and they act predominantly by permeabilising bacterial cell membranes. This mechanism is mechanistically broad — Gram-positive and Gram-negative bacteria, some fungi and viruses — and differs fundamentally from the enzymatic attacks of classical antibiotics (β-lactams at the cell wall, tetracyclines at the ribosome, fluoroquinolones at DNA topoisomerases).
From this mechanistic breadth emerged in the early 1980s a hypothesis that is still fascinating: AMPs should be effective against antibiotic-resistant bacteria because their mechanism attacks the cell membrane itself — and bacteria cannot reconstruct their membrane arbitrarily. Resistance emergence should be slow and mechanistically limited.
The first discoveries — and the first clinical programmes
Robert Lehrer in Los Angeles isolated from human neutrophils, from the early 1980s, the α-defensins HNP-1, -2, -3 — small arginine-rich cationic peptides with three disulphide bridges. Michael Zasloff discovered magainins in 1987 from the skin of the African clawed frog (Xenopus laevis). In the 1990s and 2000s more AMP families were characterised: human cathelicidins (LL-37), β-defensins, histatins in saliva, dermcidin in sweat. The library grew to hundreds of natural AMPs and thousands of synthetic analogs.
Clinical development was ambitious. A first wave from the late 1990s: iseganan (a synthetic protegrin analog) for oral mucositis, pexiganan (magainin derivative) as topical antibiotic for diabetic foot ulcers, omiganan against catheter-associated infections. Each of these substances either failed in phase 3 or made it to market but never for a systemic indication. A second wave from the 2010s (brilacidin, murepavadin, new cyclic constructs) has had similar fates.
Why this path fails — structurally
Four structural reasons are methodologically central. First: AMPs are not as effective in serum as in vitro. Albumin and other plasma proteins bind the positively charged AMPs, drastically reducing free effective concentration. What kills bacteria at 1 μM in a Petri dish barely works at 50 μM in plasma. This gap between in-vitro and in-vivo efficacy is particularly pronounced with AMPs.
Second: selectivity is limited. AMPs are designed to permeabilise bacterial membranes — but at higher concentrations they also permeabilise eukaryotic cell membranes. The therapeutic window (ratio of effective to toxic concentration) is often so narrow that no safe systemic dose can be defined. Haemolysis and renal toxicity are the most frequent limitations in phase 1 trials.
Third: resistance emergence is real. The original hypothesis — that bacteria cannot defend against membrane attacks — was too optimistic. Bacteria have developed multiple mechanisms: positively charged lipid modifications of the membrane (lipid A modifications in Gram-negatives), exo-proteases cleaving AMPs, efflux pumps for more hydrophobic AMPs. Clinical isolates show measurable AMP resistance.
Fourth: production costs are substantial. Antibiotics compete on daily doses in the cent-to-euro range. A peptide of 20-40 amino acids costs many times that per daily dose. For an indication where generic antibiotics have been available for decades, the economic justification for an expensive AMP is only possible in reserve indications for multi-resistant strains — a very narrow clinical niche.
„AMPs are wonderful in vitro. In plasma they are less wonderful. In the patient — at a dose that is not haemolytic — they are usually not effective enough. These three stages are the typical disillusionment in AMP development."
Where AMPs do work
There are indications in which the structural problems of AMPs matter less. Topical use bypasses the serum-binding and systemic toxicity problems: polymyxin B as antibiotic ointment (a cyclic lipopeptide from Bacillus polymyxa, used since the 1940s), bacitracin in wound-healing creams, daptomycin (a cyclic lipopeptide, FDA-approved 2003) for skin and soft-tissue infections. Polymyxin E (colistin) has had a renaissance as a 'last resort' intravenous antibiotic for multi-resistant Gram-negative pathogens — despite its nephrotoxicity.
These successful AMP examples are, however, predominantly natural products from bacterial fermentation, not synthetic designs from modern AMP programmes. The daptomycin story (Cubist Pharmaceuticals, 2003) is the only major success story of a modern clinical AMP — and daptomycin is administered intravenously with clinically relevant safety monitoring, comparable to other reserve antibiotics.
What the AMP story methodologically shows
Three lessons from this line are valuable for assessing new peptide hopes. First: a biologically broadly effective mechanism is not automatically a clinically usable mechanism. Membrane permeabilisation is evolutionarily ancient because it works — but it works against own cells similarly to bacteria. Selectivity arises through finely tuned membrane composition, not gross affinity. Second: the discrepancy between in-vitro and in-vivo efficacy is particularly large with AMPs. This gap should be explicitly addressed in every new peptide programme before clinical programmes begin.
Third: the economic incentive structure of antibiotic development is not peptide-friendly. Antibiotics are often used only short-term (days to weeks), often cheaply available as generics, and reserve indications have small patient populations. These realities favour expensive small molecules, not expensive peptides. The global antibiotic resistance crisis is real, but the economic solution is not found in peptide programmes.
Open questions
- Can new cyclisation and stapling strategies reduce the serum-binding and toxicity problems?
- Are AMP-antibody conjugates (AMP as targeted killing component of an antibody-drug conjugate) a productive line?
- What role do synthetic, non-natural AMP mimetics (e.g. peptoid-based structures) play in the next generation?
- What regulatory incentives (e.g. market-entry bonus for reserve antibiotics, GAIN Act in the US, EU regulation 2024) change the economic equation for AMP programmes?