PE 22-28 (mini-spadin): what the published record says

What is PE 22-28?

The molecule: a seven-amino-acid peptide, GVSWGLR, taken from the propeptide that comes off sortilin. Sortilin is also known as neurotensin receptor 3, and the propeptide is the 44-residue piece cut away by the enzyme furin when the receptor matures (Mazella, 2018).

How to read the name: PE stands for that propeptide, and the numbers are a residue range within it. So PE 12-28 is the 17-residue peptide named spadin, and PE 22-28 is the seven-residue core sitting inside it. The residue numbers are the whole identity of the molecule, which is why the naming looks like a filing reference.

The sequence, derived rather than quoted: the human sortilin entry at UniProt lists the propeptide as residues 34 to 77 of the precursor, and residues 22 to 28 of that stretch read GVSWGLR. The 2018 review names the stabilized analog G/AVSWGLR, which is the same seven letters with the opening glycine swapped for alanine, and that is the cross-check we used.

What it targets: TREK-1, a two-pore-domain potassium channel encoded by the gene kcnk2. Interest in that channel for mood began with a 2006 paper reporting that mice with the gene deleted showed a depression-resistant phenotype across five behavioral models (Heurteaux, 2006).

Where does PE 22-28 come from?

The parent: spadin, described in 2010. It bound TREK-1 with an affinity of 10 nM, blocked the channel in transfected cells and in hippocampal neurons in brain slices, raised serotonin neuron firing in the dorsal raphe, and after a 4-day intravenous course in mice raised hippocampal CREB phosphorylation and neurogenesis (Mazella, 2010).

The problem with it: it did not last. In the forced swim test the antidepressant effect of a single injection had disappeared beyond 7 hours, which the group treated as the blocker on developing it further.

The first attempt at a fix: retro-inverso analogs, built from mirror-image amino acids in reverse order. Two of them kept spadin's antidepressant effect and its neurogenesis effect, and the paper's own summary reports no side effects on pain, epilepsy processes or the cardiac level (Veyssiere, 2015). Read at the level of the individual experiment that summary is wider than the data behind it: the pain test carried both analogs, but the seizure challenge and the cardiac work were each run on only one of the two, and the cardiac work measured two repolarizing currents in transfected cells rather than a working heart. A later review from the same group names why they moved on: the analogs stretched the effect from 7 to 16 hours, and both raised in vitro toxicity concerns at higher concentrations. The 2017 paper adds that the gain was not worth what the analogs cost to synthesize (Djillani, 2017).

What produced PE 22-28: a different tactic. The group looked at how spadin breaks down in blood, took the fragments that survived, and screened them for channel activity. Shortening, rather than mirroring, is where the seven-residue peptide came from.

Is PE 22-28 the same as spadin?

No, and the difference is not only length. Spadin is PE 12-28 at seventeen residues. PE 22-28 is the seven-residue core inside it, ten residues shorter, and in the screen above it blocked more of the TREK-1 current than its parent did.

The difference that matters more: direction of effect appears to depend on how much is given. In a 2019 mouse study of focal ischemia, the same short peptide, called mini-spadin there, raised TREK-1 activity at 0.03 micrograms per kilogram and inhibited it at 3 micrograms per kilogram, a hundredfold apart. The authors used the low amount for neuroprotection in the first week and the high one for the antidepressant effect afterward (Pietri, 2019). A 2026 review in the Chinese Medical Journal singles that biphasic behavior out as a point where the short peptide parts company with spadin (Chen, 2026).

Why that is worth knowing: a molecule whose effect reverses across a hundredfold range is not one where a number can be read off a chart. This is the same class of problem as reading a mouse figure across to a person, with an extra turn in it.

Why does one molecule have four spellings?

Because no trade name was ever attached to it in anything we searched. PE 22-28, PE22-28, PE-22-28 and PE 22 28 are the same seven residues, and the variation is only how a hyphen and a space get typed around a residue range. Mini-spadin is a fifth name for the same thing, used mostly in review papers.

What that tells you about it: the laboratory coordinates stuck because nothing ever arrived to replace them. A preclinical development program did exist, and it is set out further down this page. What it never reached was the trial, the approval or the marketing decision that normally retire a filing reference. A molecule that did reach approval picked up a brand name and a label along the way. This one has neither.

Has PE 22-28 been tested in humans?

No. We found no study in which a person received PE 22-28, mini-spadin, or spadin. Searching ClinicalTrials.gov on 2 August 2026 for the query spadin returned zero studies, and the same search for mini-spadin returned zero. Running both terms, plus PE22-28 and the sequence GVSWGLR, against the World Health Organization's international registry portal on 3 August 2026 returned no results either, while the same portal returned trial records for semaglutide on that run.

What does exist in humans, and what it is not: three studies measured the natural propeptide circulating in people's blood. None of them gave anyone a peptide.

• Against healthy controls. Serum propeptide was lower in patients with major depressive disorder than in a non-psychiatric comparison group, and antidepressant treatment restored the level toward normal (Devader, 2017). The authors flag the small patient samples as the study's own limitation.
• After electroconvulsive therapy. In 45 patients with treatment-resistant major depression, serum propeptide rose between the measurement before treatment and one about a month after, and the rise reached significance only in responders. The paper's abstract adds that the rise correlated with the fall in depression rating scores. Its results section does not support that: what it reports is a correlation between propeptide concentration and the matching rating score at those two timepoints (r=-0.235, p=0.03), followed by a plain statement that no correlation was found for the change in propeptide after treatment (p=0.5). We print the results section (Roulot, 2018).
• After stroke. In a cohort of 204 stroke patients followed at 6, 12, 36 and 60 months, serum propeptide levels had a significant effect on depression scores and none on anxiety scores (Mazella, 2025).

The distinction to hold on to: those are biomarker studies of a peptide the body makes. They say something about the pathway. They say nothing about what happens when a synthetic seven-residue fragment is put into a person, and they are routinely quoted as though they do.

What has happened since 2017?

Very little that is about this fragment specifically. The work that followed went in three directions, none of which moved toward a trial, and produced a run of reviews that added no experiment at all.

• Stroke, in mice. The 2019 ischemia study above, the second and last in vivo study of the short peptide we could retrieve (Pietri, 2019).
• Pancreatic beta cells, in culture. The propeptide, spadin and mini-spadin protected beta cells against death caused by interleukin-1 beta, working through calcium and the transcription factor CREB, with mini-spadin also increasing cell proliferation (Daziano, 2021). Cells in a dish, not an animal and not a person.
• Lung cells, in culture, by somebody else. A university group unconnected to the discoverers used PE 22-28 at 7.8 micromolar to block TREK-1 in primary human alveolar epithelial cells, studying injury from high oxygen and mechanical stretch (Majumder, 2026). This is the only primary study on our list from outside the original laboratory, and it is not a test of the peptide: it uses the peptide as a reagent and cites the 2017 work for its properties.
• Reviews. Four on this list, one in 2018, two in 2019 and one in 2026. The first three come from the discovering group and restate its findings. The 2026 one is a review letter from a hospital group with no connection to that laboratory, and it summarizes the published record rather than adding experiments to it.

The development program that did exist: a national research funding record in France, for a project beginning December 2013 and running 42 months, lists a start-up and a drug delivery firm as partners alongside the discovering laboratory, carries 513,914 euros of agency support, and describes work on a long-acting spadin formulation plus adverse-effect, mutagenicity and toxicity testing "which could be proposed in a future clinical phase 1 trial" (ANR, 2013). That project concerns spadin and predates the fragment. The line that reaches the fragment is a 2019 article from the same country's research council, which names the shortened peptides, quotes the laboratory on their roughly 24-hour duration, and reports preclinical pharmacokinetic and toxicity studies underway with the same start-up, with phase 1 and phase 2 trials described as something the partners hoped a pharmaceutical company would take on (CNRS, 2019).

What that pattern amounts to: the earliest public description of the fragment we found is a March 2016 conference abstract from the same laboratory, which named it mini-spadin and reported the affinity gain a year before the full paper (Borsotto, 2016). Ten years on there is still no trial in either of the two registries we searched, and no independent group repeating the mouse antidepressant work. A funded development program did exist and did not reach a trial. Where it went after the 2019 article above is not a question the sources on this page settle, and that is a limit on what we searched rather than a finding that nothing happened. Nor is any of it evidence the molecule failed. It is an absence of the evidence that would tell you either way, and the reason a page about this molecule is short.

Is PE 22-28 an approved drug anywhere?

Not in the records we could check. Querying the openFDA drug endpoints on 2 August 2026 for spadin returned no match in the approvals database and no match in the labeling database. That is one country's record and it is a null result, so read it as what we could confirm rather than as a worldwide statement.

What that leaves: a molecule with no approval and no label in the databases we queried, which means there is no regulator's assessment of its identity or its contents to point anyone at. That is the same footing as other compounds that never finished development and kept circulating anyway, and it puts the whole burden of knowing what is in a vial onto a certificate of analysis that does not establish identity at all.

Where the evidence runs out

Six limits on what this record can tell anyone:

• One laboratory wrote nearly the whole record. Four items here set out to characterize the fragment, and all four come from the laboratory that described it: the 2016 conference abstract, the 2017 screen and mouse behavior, the 2019 stroke work and the 2021 beta-cell work. Two items sit outside that laboratory, a 2026 lung-cell paper that uses the peptide as a laboratory tool rather than studying it, and a 2026 review letter that restates published work. Neither one tests the fragment, and nobody has repeated the mouse antidepressant result.
• A paper that disagrees with itself, twice. Its abstract and its figure legend give different potency values, and its discussion swaps the two durability figures relative to its own results section and figure legend. We print both readings in both cases rather than pick a side the source does not support.
• No pharmacokinetics in people. The durability numbers are half-times of a behavioral effect in mice, not measured blood levels, and not in humans.
• No toxicology package in the published record. The safety statements are negative findings inside the discovery group's own studies. We located no repeat-dose toxicity study, no carcinogenicity study and no reproductive study. Toxicity and mutagenicity work was written into the funding record above and reported as underway in 2019, so the gap here may be an unpublished private package rather than an empty one, and either way we cannot show a reader a result.
• Direction of effect is unsettled. The 2019 study reports the channel being opened at a low amount and blocked at a high one, which means the sign of the effect is a function of quantity.
• Anti-doping status not settled by a name search. The World Anti-Doping Agency's own site would not serve its prohibited list to an automated request, so we read the 2026 list from two copies published elsewhere, one by the international testing body and one annexed to Austria's federal law gazette. Neither spadin nor mini-spadin nor any spelling of PE 22-28 appears anywhere in either. That is not a clearance: the list's first substance class covers anything it does not address in a later section that no government health authority has approved for human therapeutic use, which is wording built to catch molecules the list never names. So we make no claim about how sport governs this molecule (WADA, 2026).

Two in vivo papers on the fragment itself, both from the laboratory that described it, plus the 2016 conference abstract that previewed the first of them, in the searches set out below.

How this page is sourced

What sits behind each claim: the pharmacology and the mouse behavior come from the 2017 primary paper plus the five earlier studies it builds on, the stroke, beta-cell and lung-cell findings from three later primary papers, the naming and the sequence cross-check from four reviews and one protein database entry, the earliest public description from a 2016 conference abstract, the human biomarker findings from three clinical studies, the development history from a public funding record and a research council article, and the absence claims from registry, regulator and literature queries we ran and dated.

Twenty-four sources: fifteen journal articles carrying both a linked DOI and a PubMed ID, one conference abstract carrying a DOI and no PubMed record, one 2026 review letter reachable through its PubMed record after the publisher's host refused an automated request, one protein database entry, two trial registry searches, one regulator database query, one national research funding record, one research council article retrieved from a web archive, and one anti-doping list read from copies published elsewhere because the issuing agency's own host would not serve it. That is a small list, and it is a record of what these searches returned rather than a claim to have seen everything ever published on this molecule.

How the count was tested, and what the test actually was: on 2 August 2026 we typed PE 22-28 into PubMed without quotation marks. That returns 17 records and we read every one. Sixteen are unrelated records where the string is a journal page range or some other numeric span, covering prostate cancer screening, ulcerative colitis, snail parasitology and influenza serology. The seventeenth is the 2017 paper below. We then searched spadin across all fields, which returns 80 records, and read the title and abstract of every one: 40 name spadin, of which one is the 2017 paper, one is the 2019 stroke study using the short peptide, and the rest concern spadin itself, the parent propeptide, or TREK-1 in other tissues.

Why that 17 is not what it looks like, which matters more than the number: the unquoted query is not a phrase search. PubMed reports back exactly what it did with it, translating it to ("popul econ"[Journal] OR "philos explor"[Journal] OR "palaeontol electronica"[Journal] OR "pe"[All Fields]) AND "22-28"[All Fields]. It read PE as a set of journal abbreviations plus the bare letters, and paired that with the numeric span. That is why sixteen page ranges came back. The phrase search proper, "PE 22-28"[All Fields], returns exactly 1 record, the 2017 paper. So the sixteen were retrieved by an accident of query parsing, and the honest description of the seventeen is that we read whatever a loose query returned, not that a phrase search found seventeen things.

What none of those queries can see: PubMed indexes titles and abstracts, so a paper that uses this peptide in its methods and never names it in the abstract is invisible to every search above. Running the name against a full-text index instead, Europe PMC on the same date, returns the 2026 lung-cell paper on this list, which names the peptide in its methods, results and figure captions and nowhere in its abstract. One database found it and the other structurally could not. There is a second hole in the same shape, and it cost us a date. Neither index carries the March 2016 conference meeting abstract in which this laboratory first named mini-spadin: a Crossref title query for mini-spadin on 3 August 2026 returns that record at the top, and asking PubMed for its exact title returns nothing, while the same PubMed route returns the 2017 paper when asked for its exact phrase. A name search sees what its index holds, never what has been published. That is the reason this page does not claim a complete literature: what these searches support is a statement about named queries on named databases on one date, and an absence claim is only ever as good as the search behind it.

On the sequence: we did not find the seven letters printed as a sequence in the 2017 paper. We derived GVSWGLR by taking residues 22 to 28 of the propeptide as UniProt defines it, and checked the result against the 2018 review, which names the substituted analog G/AVSWGLR. Two independent routes to the same string is why we print it.

The standard: how we verify a citation, and who reads a page before it goes out, is written up on the methodology page. A thin literature is exactly where that standard earns its keep.

Last reviewed: 2 August 2026.

1. 1

   Djillani A, Pietri M, Moreno S, Heurteaux C, Mazella J, Borsotto M. Shortened spadin analogs display better TREK-1 inhibition, in vivo stability and antidepressant activity. Front Pharmacol. 2017;8:643. doi:10.3389/fphar.2017.00643. PMID 28955242. This is the primary source for PE 22-28 and the only paper we found reporting the fragment's own screen, potency, selectivity and mouse behavior.

2. 2

   Mazella J, Pétrault O, Lucas G, et al. Spadin, a sortilin-derived peptide, targeting rodent TREK-1 channels: a new concept in the antidepressant drug design. PLoS Biol. 2010;8(4):e1000355. doi:10.1371/journal.pbio.1000355. PMID 20405001.

3. 3

   Heurteaux C, Lucas G, Guy N, et al. Deletion of the background potassium channel TREK-1 results in a depression-resistant phenotype. Nat Neurosci. 2006;9(9):1134-1141. doi:10.1038/nn1749. PMID 16906152.

4. 4

   Veyssiere J, Moha Ou Maati H, Mazella J, et al. Retroinverso analogs of spadin display increased antidepressant effects. Psychopharmacology (Berl). 2015;232(3):561-574. doi:10.1007/s00213-014-3683-2. PMID 25080852.

5. 5

   Moha Ou Maati H, Veyssiere J, Labbal F, et al. Spadin as a new antidepressant: absence of TREK-1-related side effects. Neuropharmacology. 2012;62(1):278-288. doi:10.1016/j.neuropharm.2011.07.019. PMID 21807005. The source of this page's spadin safety findings on pain, kainate seizures, infarct size after focal ischemia, blood glucose and the cardiac measures. It tested spadin, not the shortened fragment. Two other entries carry the rest of the safety picture: the analog-specific pain, seizure and cardiac-current work is reference 4, and the fragment's own selectivity against four related channels and its hERG result are reference 1.

6. 6

   Moha ou Maati H, Peyronnet R, Devader C, et al. A human TREK-1/HEK cell line: a highly efficient screening tool for drug development in neurological diseases. PLoS One. 2011;6(10):e25602. doi:10.1371/journal.pone.0025602. PMID 22022421. The cell line in which every potency figure on this page was measured.

7. 7

   Pietri M, Djillani A, Mazella J, Borsotto M, Heurteaux C. First evidence of protective effects on stroke recovery and post-stroke depression induced by sortilin-derived peptides. Neuropharmacology. 2019;158:107715. doi:10.1016/j.neuropharm.2019.107715. PMID 31325429. Source of the biphasic finding and of both amounts quoted for it.

8. 8

   Mazella J, Borsotto M, Heurteaux C. The involvement of sortilin/NTSR3 in depression as the progenitor of spadin and its role in the membrane expression of TREK-1. Front Pharmacol. 2018;9:1541. doi:10.3389/fphar.2018.01541. PMID 30670975. The review that names the sequence 22 to 28 as mini-spadin and prints the substituted analog G/AVSWGLR.

9. 9

   Djillani A, Mazella J, Heurteaux C, Borsotto M. Role of TREK-1 in health and disease, focus on the central nervous system. Front Pharmacol. 2019;10:379. doi:10.3389/fphar.2019.00379. PMID 31031627. Source of the 7 to 16 hour figure for the retro-inverso analogs and of the in vitro toxicity concern that ended that line of work.

10. 10

    Djillani A, Pietri M, Mazella J, Heurteaux C, Borsotto M. Fighting against depression with TREK-1 blockers: past and future. A focus on spadin. Pharmacol Ther. 2019;194:185-198. doi:10.1016/j.pharmthera.2018.10.003. PMID 30291907.

11. 11

    Daziano G, Blondeau N, Béraud-Dufour S, et al. Sortilin-derived peptides promote pancreatic beta-cell survival through CREB signaling pathway. Pharmacol Res. 2021;167:105539. doi:10.1016/j.phrs.2021.105539. PMID 33737242.

12. 12

    Roulot M, Minelli A, Bortolomasi M, et al. Increased serum levels of sortilin-derived propeptide after electroconvulsive therapy in treatment-resistant depressed patients. Neuropsychiatr Dis Treat. 2018;14:2307-2312. doi:10.2147/NDT.S170165. PMID 30233189. Measures the endogenous propeptide. Nobody in it received a synthetic peptide.

13. 13

    Mazella E, Mendyk AM, Accart B, et al. Serum sortilin-derived propeptide concentrations as markers of depression in chronic stroke. J Neurol Sci. 2025;472:123459. doi:10.1016/j.jns.2025.123459. PMID 40107034. Same distinction: an observational biomarker study, not an administration study.

14. 14

    Chen M, Wang Y, Xiong Z, Fang Y, Zhu Z. TREK-1 in central nervous system diseases. Chin Med J (Engl). 2026;139(4):620-622. PMID 40539288. doi:10.1097/CM9.0000000000003662. Retrieved 2 August 2026 via PubMed; the publisher's own host returned HTTP 403 to an automated request for the DOI, so the PubMed record is the locator we can vouch for.

15. 15

    UniProt Consortium. SORT_HUMAN, sortilin, gene name SORT1, accession Q99523, feature PROPEP residues 34 to 77. Retrieved 2 August 2026, HTTP 200. rest.uniprot.org, Q99523. The propeptide boundaries used to derive the seven-residue sequence.

16. 16

    ClinicalTrials.gov, U.S. National Library of Medicine. Registry searched 2 August 2026 through the version 2 API, query term spadin and query term mini-spadin. Both returned an empty study list. clinicaltrials.gov, spadin query. The same endpoint queried for semaglutide on 3 August 2026 returned 757 studies, which is the control showing it answers when there is something to find.

17. 17

    U.S. Food and Drug Administration, openFDA. Drugs@FDA endpoint queried on generic name spadin and drug labeling endpoint queried on spadin, both on 2 August 2026, both returning NOT_FOUND. api.fda.gov, Drugs@FDA query. The same query run for semaglutide returns one record, which is the control showing the endpoint answers when there is something to find.

18. 18

    Devader C, Roulot M, Moréno S, et al. Serum sortilin-derived propeptides concentrations are decreased in major depressive disorder patients. J Affect Disord. 2017;208:443-447. doi:10.1016/j.jad.2016.10.049. PMID 27838145. The third human serum study, and like the other two it measures the endogenous propeptide rather than administering anything.

19. 19

    Majumder N, Lopez B, Kasparian A, et al. Role of the two-pore domain potassium channel TREK-1 in hyperoxia- and mechanical stretch-induced alveolar epithelial injury. Am J Physiol Lung Cell Mol Physiol. 2026;330(1):L49-L65. doi:10.1152/ajplung.00143.2025. PMID 41252294. Full text read at PMC12716432, because the publisher's own host returned HTTP 403 to an automated request for the DOI. One of two items on this list from outside the discovering laboratory, and the only primary study among them. Its methods name PE22-28 at 7.8 micromolar as the TREK-1 blocker used; the peptide appears nowhere in its abstract, which is why a title and abstract search cannot find it.

20. 20

    Agence Nationale de la Recherche. Project ANR-13-RPIB-0002, "Medication of a new concept on depression". Project summary listing partners, 513,914 euros of agency support, and a start date of December 2013 for 42 months. Retrieved 2 August 2026, HTTP 200. anr.fr, ANR-13-RPIB-0002. Source of the quoted phrase about a future clinical phase 1 trial. A request for a non-existent project identifier on the same host returns HTTP 404, which is the control showing this 200 is a real record.

21. 21

    Centre National de la Recherche Scientifique. "Des minipeptides pour contrer la dépression", CNRS Innovation newsletter item 653, dated 18 April 2019. Retrieved 2 August 2026 from an Internet Archive snapshot, HTTP 200, because the live page serves no content to an automated request. web.archive.org, CNRS newsletter 653. Names the shortened peptides, and reports preclinical pharmacokinetic and toxicity studies underway with the start-up partner.

22. 22

    World Anti-Doping Agency. Prohibited List 2026, World Anti-Doping Code International Standard, in effect 1 January 2026. Read 3 August 2026 from the copy published by the International Testing Agency, which served the full 26-page document. ita.sport, 2026 Prohibited List. The issuing agency's own host answered the same request with HTTP 202, a zero-byte body and the header x-amzn-waf-action set to challenge, so this is the locator we can vouch for. The same text is also annexed to Austria's federal law gazette of 30 December 2025, and we read that second copy as a cross-check: the two agree. Searched in both for spadin, mini-spadin, PE 22-28, PE22-28, PE-22-28 and GVSWGLR, with zero occurrences of any of them. BPC-157, which this same document does name, is the control showing the search itself works.

23. 23

    Borsotto M, Djillani A, Devader C, Heurteaux C, Mazella J. The Mini-Spadin, an efficient alternate to Spadin in the depression treatment. Eur Psychiatry. 2016;33(S1):S407. doi:10.1016/j.eurpsy.2016.01.1470. A conference meeting abstract, not a full paper, and the earliest public description of the fragment we found. It carries no PubMed record, which is why every search above missed it: we reached it through a Crossref title query on 3 August 2026, and the DOI then resolved to the publisher's article page carrying the same title and abstract. It reports the 400-fold affinity gain and the same two behavioral tests the 2017 paper reports in full, so we read it as a preview of that dataset rather than as a separate experiment.

24. 24

    World Health Organization, International Clinical Trials Registry Platform. Search portal queried 3 August 2026 for spadin, mini-spadin, PE22-28 and GVSWGLR. Every one returned a page reading "No results were found". trialsearch.who.int, spadin query. The same portal queried for semaglutide on the same date returned trial records, which is the control showing it answers when there is something to find.

Related pages

Around this page: five neighboring pages on thin evidence, on what a compound is, and on what a test result covers.

• Cardarine (GW501516) explained: another compound whose development stopped and whose retail life carried on, with a much heavier document trail.
• MOTS-c guide: the same mouse-to-human problem, on a peptide with more human data than this one.
• What an HPLC purity number misses: the reason a high percentage on a certificate still leaves the identity question open.
• What a lyophilized peptide is: what freeze-drying does to a peptide, and what the powder it leaves behind does and does not tell you.
• PT-141 and the Vyleesi approval: the opposite case, a peptide that did reach approval, and exactly what that approval covers.
• FOXO4-DRI, the senolytic: another compound whose entire administered record is preclinical, read the same way, with the amounts kept in the animals they came from.

Next in the series: we open one molecule's paper trail a week and send it out in The Decadewise briefing.