Hexapeptide-9 has one of the smallest published files of any peptide sold in beauty: three records, counted on 2026-09-19 across both hyphenations. What makes it worth an entry is not the size of the file but its shape. Two of the three papers exist because of the same problem, and the third is about something else.
The molecule
| Sequence | Gly-Pro-Gln-Gly-Pro-Gln |
| Systematic name | glycyl-L-prolyl-L-glutaminylglycyl-L-prolyl-L-glutamine |
| PubChem CID | 16129319 |
| Formula / weight | C24H38N8O9 / 582.6 g/mol |
| CAS | 885024-94-2 |
| Trade name | Collaxyl IS |
The sequence is two repeats of glycine-proline-glutamine. Glycine followed by proline is the motif that repeats through collagen itself, and the authors of its clinical trial describe the material as a collagen peptide. At 582.6 g/mol it is heavier than the copper tripeptide and lighter than the palmitoyl peptides that dominate the shelf.
A search for the trade name Collaxyl in titles and abstracts returns zero records. So does a search for the sequence GPQGPQ. Everything published about this ingredient is published under its INCI name.
The three records
1. The trial, and it tested a ring rather than a chain. A randomized, double-blinded, active- and vehicle-controlled study assigned healthy volunteers with ageing skin to twice-daily 0.002% cyclized hexapeptide-9 serum, 0.002% retinol serum, or vehicle, for 56 days.
| Endpoint | Cyclized hexapeptide-9 | Retinol 0.002% |
|---|---|---|
| Crow's feet — number | −2.20 (95% CI −4.38, −0.03) | not significant |
| Crow's feet — area | −3.95 (−5.80, −2.11) | −2.23 (−3.86, −0.60) |
| Crow's feet — roughness | −1.95 (−3.30, −0.59) | not significant |
| Forehead — number | −2.88 (−4.21, −1.56) | −1.05 (−1.69, −0.41) |
| Forehead — area | −4.90 (−5.97, −3.82) | reported as decreased |
| Forehead — roughness | −3.96 (−5.92, −2.01) | not significant |
Except for crow's-feet roughness, the peptide arm showed significantly larger effects than the retinol arm on every outcome, and the authors report a time-dependent increase in potency with longer use. Three things bound that result. The abstract does not state the number of volunteers. The retinol comparator sits at 0.002%, which is a low concentration for retinol. And the tested molecule is not the one on a label.
2. Why the molecule was modified at all, in the trial's own words. The study's background states that topical peptide use is "limited due to low stability and poor skin permeability", and that the cyclized form was developed with "increased stability and skin permeability". Cyclization joins the ends of the peptide into a ring. It is a structural change, made because the linear molecule is not good enough at the job — and INCI nomenclature has no way to declare it, so both forms are simply "hexapeptide-9" on a carton if they appear there at all.
3. The second paper solves the same problem from the other end. Rather than changing the peptide, it changes the skin. An ionic liquid prepared from gamma-aminobutyric acid and lactic acid was used as a penetration enhancer, raising cumulative permeation of hexapeptide-9 4.79-fold and subcutaneous retention 7.89-fold over twelve hours against no enhancer. Its opening premise is that peptides "exhibit poor transdermal efficiency". That is two independent research groups, in the same year, building different machinery around the same limitation.
4. The third record is a different ingredient. A 2026 framework paper on evaluating the safety of cosmetic peptides with bioinformatic tools runs a panel through BLASTp, ToxinPred, Peptipedia, BIOPEP-UWM, AllerCatPro and IEDB. The peptide in that panel is caffeoyl hexapeptide-9, a caffeic-acid conjugate with its own INCI name. It is a useful paper about a useful method, and it is not evidence about this molecule.
What the file adds up to
Strip the three records down and the position is unusually clean for a cosmetic ingredient:
- Zero published studies test hexapeptide-9, as declared, applied to people.
- One clinical trial tests a cyclized derivative at 0.002%, and it outperformed retinol at the same concentration on most wrinkle endpoints.
- Two of the three records are penetration engineering, which is to say that the ingredient's own literature treats getting it into skin as the unsolved part.
That last point is the one worth carrying to a shelf. The general question of whether peptides cross the stratum corneum is the standing argument in this category, and most ingredients leave it implicit. Here the researchers state it outright, twice, and then spend their papers working around it.
Our entries on hexapeptide-11 and palmitoyl tripeptide-1 set out how much deeper a file can be; the ingredient-list guide covers what a declaration can and cannot tell you about which form of a molecule you have, and the peptide-versus-retinoid count puts the 0.002% head-to-head in its wider context — nine randomized trials for topical peptides against fifty-five for retinoids.
What would change this page
A trial of linear hexapeptide-9 as declared, at a stated concentration, against vehicle. Or the cyclized trial's full text with its participant count, which would let the effect sizes above be weighed properly. Both are ordinary requests; neither is currently answerable from the published record.
Sources
- Chang H, Tao K, Yang Y, et al. Novel cyclized hexapeptide-9 outperforms retinol against skin aging: a randomized, double-blinded, active- and vehicle-controlled clinical trial. J Cosmet Dermatol. 2025. PMID 40586182.
- Pioneering ionic liquids in neuro-soothing: enhanced transdermal delivery of collagen peptides and their synergistic anti-aging functions. Mater Today Bio. 2025. PMID 39968525.
- A framework for the safety evaluation of peptides in cosmetics. Curr Res Toxicol. 2026. PMID 41953401.
- PubChem compound record, CID 16129319 (hexapeptide-9), retrieved 2026-09-19.
- Record counts: NCBI PubMed E-utilities, searched 2026-09-19, hyphenated and unhyphenated forms plus trade name and sequence.