pe-22-28

seven amino acids off the sortilin propeptide. three papers, one lab, zero humans, and a duration number that belongs to a different molecule.

tier C · focus · 3 papers · zero humans

verdict

a seven-residue TREK-1 blocker with genuine dose-response electrophysiology and repeating rodent behaviour, all of it from one CNRS lab, with no human exposure and no published pharmacokinetics.

on whether PE 22-28 does anything in people — nobody knows, because nobody has been given it. a july 2026 sweep of ClinicalTrials.gov returned zero studies for spadin, zero for PE 22-28 and zero for mini-spadin, and a PubMed search restricted to clinical-trial publication types returned nothing either. the three human papers vendors point at measured the body's own sortilin propeptide in serum samples, with nothing administered to anyone. the entire human file is a blood test.

on the 23 hour duration on the label — that number belongs to a different molecule. the origin paper measured four half-effect times, 14, 17, 21 and 23 hours, and every one of them was measured on the alanine-substituted analog G/A-PE 22-28 or on its biotinylated version. plain PE 22-28 was never in the duration experiment. the paper reports no serum half-life number for it anywhere, and no pharmacokinetic study of the molecule has been published since.

on whether the TREK-1 mechanism holds up — the target holds up better than the molecule. independent groups in nanjing, dalian, budapest and seoul have worked on TREK-1 blockade, and a 2025 mouse paper identified the channel as a target of the ketamine metabolite hydroxynorketamine. the blocking step is contested. a group at portsmouth applied 1 micromolar spadin to arachidonic-acid-activated TREK-1 in xenopus oocytes and saw no inhibition, on the mouse clone and on the human clone, at a concentration their own paper says should give about 75 percent block.

based on published preclinical evidence. research framing only. not medical advice.

why C-tier

C-tier, and the ceiling is fixed before the argument starts: no human has ever received this peptide, so B is unreachable however clean the mouse work reads. It lands on C rather than D because the preclinical work is genuine rather than decorative. The target is defined, the sequence is confirmed independently by the propeptide recited in a granted patent and by that patent's own sequence listing, the electrophysiology is dose-response rather than single-point, and the behavioural readouts repeat across four paradigms at 3 to 4 mcg/kg with large effect sizes. The published record also carries a real internal negative control: PE 22-27, the fragment that failed to block the channel, is also reported ineffective in forced swim. What holds it at the bottom of C: every paper naming this exact molecule comes from one CNRS group, group sizes run five to ten animals, there is no pharmacokinetics, no toxicology and no in vivo dose ranging on the plain peptide, one independent laboratory could not reproduce channel block on the parent peptide, the granted patents deliberately claim other molecules, and the reported direction of effect on TREK-1 reverses across a hundred-fold dose range with no mechanistic account. That is a low C. If a second laboratory publishes original data on this peptide, or a first-in-human study registers, the tier moves.

the core tension

the molecule is real and the target is credible, and almost every number sold with it was measured on something else. the 23 hour duration belongs to the biotinylated alanine analog at 40 mcg/kg. the granted US patent claims AVSWGLR and leaves the GVSWGLR being sold pointedly unclaimed, with the specification saying why: everything except the alanine analog is a natural product. the potency multiplier is 333-fold in the patent, more than 300-fold in the origin paper, 40 to 350-fold in the lab's own review, and 300 to 500 on vendor pages. the one independent group that tried to reproduce channel block on the parent peptide could not. and the reported direction of effect on TREK-1 flips between 0.03 and 3 mcg/kg with no mechanistic account of why.

what it is

PE 22-28 is a seven-residue fragment of the 44-amino-acid propeptide that sortilin sheds when it matures. The propeptide sequence recited in the granted CNRS patent is QDRLDAPPPPAAPLPRWSGPIGVSWGLRAAAAGGAFPRGGRWRR, and positions 22 to 28 of it read Gly-Val-Ser-Trp-Gly-Leu-Arg. The parent molecule is spadin, the longer 12-28 fragment the same lab described in PLoS Biology in 2010. PE 22-28 is the shortened version, and the lab's own name for it in print is mini-spadin.

what it does

In cells stably expressing human TREK-1, PE 22-28 blocked 55 percent of the arachidonic-acid-activated current at 100 nanomolar, across 13 recordings, at a reported IC50 of 0.10 to 0.12 nanomolar. TREK-1 is a two-pore potassium channel, and blocking it produces antidepressant-like behaviour in rodents in work from several independent groups. In mice, a single 3.0 mcg/kg intraperitoneal dose cut forced-swim immobility from 161.7 seconds to 91.8 seconds, ten animals per arm. Hippocampal BrdU-positive cell counts came in at 1736 against 899 for saline, five animals per arm.

origin

Institut de Pharmacologie Moleculaire et Cellulaire, CNRS, at Valbonne near Nice. Jean Mazella, Marc Borsotto and Catherine Heurteaux described spadin in PLoS Biology in 2010 and the shortened analogs in Frontiers in Pharmacology in 2017. On an exact-string PubMed search, three records name this molecule: the 2017 origin paper, a 2019 stroke paper and a 2021 beta-cell paper. All three come from that group. Outside reviews mention it. None of them generate data on it.

why researchers are interested

The pitch is potency and duration. Vendor pages describe a compound 300 to 500 times stronger than spadin that lasts about 23 hours, sold into the mood and nootropic corner of the market. Both numbers trace back to the same 2017 paper, and both were measured on modified versions of the molecule rather than on the one being sold.

does it work

In mice the behavioural readouts repeat across three papers and the effect sizes are large. In humans the question has never been asked. Group sizes run five to ten. There is no published pharmacokinetics, no toxicology and no in vivo dose ranging on the plain peptide, which was only ever given at 3.0 to 4.0 mcg/kg by injection and once by gavage at 1 mg/kg. And the direction of effect on the target reverses across a hundred-fold dose range, which the papers report and do not explain.

claims vs the data

• 300 to 500 times more potent than spadin — partially true — The published multipliers are "more than 300 fold" in the origin paper, 333-fold in the granted patent, and 40 to 350 times in the lab's own 2019 review. The 500 does not appear in any primary document. The denominator moves too: the 2017 paper cites spadin at 40 to 60 nM, while the 2010 discovery paper's own measured IC50 was 70.7 nM.
• lasts about 23 hours — contradicted — The origin paper reports four half-effect times, 14 and 21 hours for G/A-PE 22-28 at 3.2 and 32 mcg/kg, and 17 and 23 hours for the biotinylated analog at 4.0 and 40 mcg/kg. Plain PE 22-28 was not dosed in that experiment. The paper's own discussion then reverses two of those values against its results section, and reports no serum half-life number for the peptide anywhere.
• patented technology — contradicted — US 11,220,527 B2 claim 1 covers the peptide G/A-PE(22-28), sequence AVSWGLR (SEQ ID NO: 7), and its biotinylated form. GVSWGLR is SEQ ID NO: 3, appears in the specification only, and is claimed nowhere. The specification gives the reason: the peptides of the invention, with the exception of G/A-PE(22-28), are natural products. The earlier family patent, US 10,251,937 B2, claims only end-capped retro-inverso D-peptides.
• a nootropic that sharpens cognition and memory — unsupported — No learning, memory or attention task in an intact healthy animal has been published on this peptide. The only cognitive readout in the file is the 2019 stroke paper, where mini-spadin improved motor and cognitive deficits after focal cerebral ischaemia. Rescuing an injury-induced deficit in a mouse is a different claim from enhancement in a healthy one.
• investigational human range is 0.5 to 2 mg subcutaneous — contradicted — There is no investigational human range, because there has been no investigation in humans: no IND located, no registered trial, no published exposure. Published rodent dosing runs 0.03 to 100 mcg/kg by injection and 1 mg/kg by gavage. Scale check from this review's own arithmetic: at the published 3 mcg/kg mouse dose a 25 g animal receives 0.075 mcg, so a single 10 mg vendor vial holds roughly 133,000 such doses.
• human studies show it helps depression — contradicted — The three human papers are observational blood-draw studies of the body's own 44-residue propeptide, roughly 250 to 300 subjects across partly overlapping author groups. Nothing was administered to anyone. The two major-depression results also point in different directions: the 2017 paper reports lower serum propeptide in patients than controls at p=0.035, and the 2018 study finds no baseline difference at all, t=0.10, p=0.92.
• blocks TREK-1 at 0.1 nanomolar — partially true — The origin paper gives both 0.10 and 0.12 nM for this peptide in different places within the same paper, and the granted patent assigns 0.12 nM to the alanine analog instead and gives no figure for the plain peptide at all. The assay also pre-activates the channel with 10 micromolar arachidonic acid, and the paper concedes weak direct inhibition of TREK-1 under basal conditions, so the block is state-dependent.
• independent labs have confirmed it — partially true — TREK-1 as an antidepressant target has genuine outside support: Nanjing, Dalian, Budapest, Seoul and Portsmouth have all published on it, and a 2025 mouse study identified the channel as a target of hydroxynorketamine. Every paper on this heptapeptide comes from the originating group. The Portsmouth group, testing the parent peptide, could not reproduce channel block on the mouse or human clone.
• devoid of side effects — overreach — The origin paper's own sentence is conditional, an expectation extrapolated from spadin data rather than a finding, and the granted patent's broader version of the claim carries no supporting data at all. No toxicology study of PE 22-28 has been published. The lab's 2011 paper reports glycaemia unchanged by spadin; its 2016 paper reports spadin roughly doubling plasma insulin and lowering blood glucose in mice under a section heading naming hypoglycaemia.

key facts

• molecular formula: C35H55N11O9
• molecular weight: 773.9 g/mol (calculated from the sequence; the patent arithmetic in this review used about 774)
• amino acids: 7
• half-life: not published. no pharmacokinetic study of PE 22-28 exists, and the origin paper reports no serum half-life number for it. the 23 h figure quoted by vendors is a behavioural half-effect time measured on the biotinylated alanine analog at 40 mcg/kg
• type: heptapeptide; fragment 22-28 of the sortilin (NTSR3) propeptide; TREK-1 blocker
• CAS: 1801959-12-5 circulates in vendor and catalogue listings; not confirmed against a primary registry in this review, and PubMed does not resolve it
• 3 papers naming this exact molecule
• 0 humans given PE 22-28 or spadin
• 5 mice per group in the neurogenesis result
• 1 laboratory behind every original paper

frequently asked questions

What is PE 22-28?

PE 22-28 is a seven-amino-acid peptide, Gly-Val-Ser-Trp-Gly-Leu-Arg, corresponding to positions 22 to 28 of the propeptide that human sortilin sheds during maturation. It is a shortened version of spadin, and the originating laboratory calls it mini-spadin. It is sold as a research chemical and has no approved human use anywhere.

What does PE 22-28 do?

In a stable human TREK-1 cell line it blocks that potassium channel, inhibiting 55 percent of the arachidonic-acid-activated current at 100 nanomolar, at a reported IC50 of 0.10 to 0.12 nanomolar. In mice, 3.0 mcg/kg intraperitoneal reduced immobility in forced swim, reduced latency in novelty-suppressed feeding, shortened escape latencies in learned helplessness, and roughly doubled hippocampal BrdU-positive cell counts. A 2019 paper reports improved motor and cognitive deficits in a mouse stroke model. Every one of these results is preclinical and all of them come from one research group.

How is PE 22-28 administered in the published research?

In the literature it was given to mice by intraperitoneal injection at 3.0 to 4.0 mcg/kg, with activity also reported from 0.03 mcg/kg in a stroke model, and the shortened analogs stayed active when given by gavage at 1 mg/kg. There is no human administration protocol, because there is no human study. A july 2026 registry sweep returned zero studies for PE 22-28, mini-spadin or spadin.

Does PE 22-28 last 23 hours?

Not on the published record. The 23 hour figure is a half-effect time measured on the biotinylated alanine analog at 40 mcg/kg. All four values the origin paper reports, 14, 17, 21 and 23 hours, were measured on G/A-PE 22-28 or its biotinylated form. Plain PE 22-28 was never included in that experiment, the paper reports no serum half-life number for it, and no pharmacokinetic study of it has been published.

What are the side effects of PE 22-28?

Uncharacterised. No toxicology study, no pharmacokinetic study and no human exposure exist for this peptide. The often-quoted line about being free of side effects is conditional in the original text, applies to spadin analogs generally, and rests on a battery of negative findings in mice rather than a toxicology package. The same laboratory's own record on glucose does not agree with itself: a 2011 paper reports glycaemia unchanged by spadin, and a 2016 paper reports that spadin at 8 mcg/kg intravenous roughly doubled plasma insulin in mice and cut the glucose area under the curve by 28 percent, under a section heading naming hypoglycaemia.

Is PE 22-28 FDA approved?

No. It has no FDA approval, no IND located on the public record, no EMA record and no registered clinical trial in any registry searched. It is sold as a research chemical. The two registry numbers that circulate alongside it, NCT01330160 and NCT06943365, are both observational and neither administers any compound.

related peptides

• dihexa — the other single-lab rodent neuro compound whose marketing numbers come from a related molecule
• humanin — same shape of human file: observational biomarker measurement, zero interventional trials
• semax — nootropic peptide with an actual human trial base, for the contrast
• selank — anxiolytic research peptide with a clinical record, thin but real

reptides grades the research record and cites the literature behind every call. research reference only; not medical advice.