What it is
Melittin is 26 amino acids long — GIGAVLKVLTTGLPALISWIKRKRQQ, with an amide cap at the end. The western honeybee makes it in its venom gland as a 70-residue precursor (UniProt P01501) and trims it to the working peptide.
It dominates the venom. A 2024 review (Jafari et al., Archiv der Pharmazie, PMID 38251938) puts it at roughly 40–60% of bee venom's dry weight, notes that it is the main index by which venom is priced, and names it as the cause of the pain. The rest of the venom — apamin, mast-cell degranulating peptide, the enzyme phospholipase A2, histamine — makes up the remainder.
What it does to a cell
Look at the sequence and the design is visible. The first twenty residues are mostly water-repelling (isoleucine, valine, leucine, alanine); the end carries a cluster of positive charges (K-R-K-R). Folded into a helix, one face is oily and the other charged.
A molecule shaped like that slides into the fatty double layer of a cell membrane and, in numbers, opens pores in it. The cell leaks and dies. It does this to bacteria, to viruses with fatty envelopes, to tumour cells in a dish — and to red blood cells. Breaking red cells open is called haemolysis, and melittin is one of the best-known haemolytic peptides.
That one mechanism explains both halves of its reputation.
The promise, and what holds it back
PubMed returns 3,095 records for melittin (2026-10-28) — a large literature for a single toxin, much of it on cancer cells, antimicrobial action and drug delivery. The 2024 review summarises the field's central problem in plain terms: "the clinical application of MLT has been limited by its severe hemolytic activity." Most recent work tries to fix that by changing the sequence, expressing it in engineered cells, or loading it into nanoparticles that release it only near a target.
The human record for melittin as a treatment is close to empty:
| Record | What it tested | Melittin's role |
|---|---|---|
| Koyama 2000, Pain (PMID 10666517) | Whether melittin can serve as an experimental pain agent in people; 5 μg injected into forearm skin | Used as a pain stimulus: "temporarily produced severe pain", and 10% lidocaine gel did not significantly blunt it |
| Barbosa 2021, Front Immunol (PMID 33841437) | An antivenom for mass bee stings, 20 people, single arm | Measured as venom in blood; detected in 8 people at day 30 |
| Hartmann 2016 (PMID 27808347) | Melittin for FIV-infected cats, 10 vs 10 placebo | Treatment — no significant efficacy |
| NCT02535416, NCT02363946 | Arrowhead's ARC-520 / ARC-AAT RNA medicines | A melittin-like peptide in the delivery vehicle, measured as an analyte |
The last row is the closest melittin has come to the clinic: as a membrane-disrupting helper that gets an RNA drug out of the compartment it enters a cell through. Neither study was a test of melittin's own effect.
Whole venom is not melittin
Bee venom acupuncture and venom-based creams have their own small trial literature — PubMed tags 31 bee-venom records as randomised trials. Those test the whole mixture. A result from one of them cannot be credited to melittin, and a sting is a dose of every component at once. Other membrane-active peptides in this library did reach approval, as antibiotics — colistin and daptomycin — which makes them the useful contrast: melittin's target is any membrane, including ours.
Where it stands
Approved as a medicine: nowhere, as melittin.
Human evidence as a treatment: none published.
Practical relevance: the pain and toxicity of bee stings. Anyone with a history of strong reactions to stings needs an allergist's advice; a severe reaction after a sting is an emergency.
Last checked
2026-10-28.