FOXO4-DRI

Chemical identity, and two records under one name

The peptide is forty-six residues long. Baar and colleagues print it in their 2017 methods in lower case, the convention for D-configuration: H-ltlrkepaseiaqsileaysqngwanrrsggkrppprrrqrrkkrg-OH, molecular weight 5358.2, synthesised to greater than 95 percent purity. The last twelve residues, read in reverse, are the HIV-1 TAT cell-penetrating motif GRKKRRQRRR (TAT 48-57) followed by a two-proline linker; this is the segment Bourgeois and colleagues call HIV-TAT. The preceding thirty-four, PRKGGSRRNAWGNQSYAELISQAIESAPEKRLTL, correspond exactly to residues 91 to 124 of human FOXO4 (UniProt P98177). Retro-inverso construction reverses the residue order and inverts the chirality of every centre, which approximately reproduces the side-chain topology of the parent L-peptide while removing the substrate geometry that mammalian proteases recognise.

PubChem returns two compound records under the name FOXO4-DRI. CID 167312269 and CID 168431240 carry the same molecular formula, C228H388N86O64, the same average mass of 5358, and the same CAS number in their synonym lists, but their InChIKeys differ in the stereochemical block. The two records differ only in the C-terminal cationic segment. Reading each residue in context in the isomeric SMILES, CID 167312269 writes the C-terminal lysines and arginines as N[C@H](R)C(=O), the D configuration, which matches the all-D peptide in the 2017 methods; CID 168431240 writes them as N[C@@H](R)C(=O), the L configuration. Both records carry a mixture of @ and @@ tags overall, thirteen and thirty-two in the first and twenty-one and twenty-two in the second, because isoleucine and threonine each contribute a second stereocentre, so a raw tag count does not distinguish them.

That sequence contains no cysteine and no methionine, so it contains no sulfur. A molecular formula circulating on aggregator pages, C234H380N76O68S2, with a mass of 5382.08 and a residue count of forty-eight, fails on all three counts. Computing the formula from the published sequence returns C228H388N86O64 and 5358.1, which is what both PubChem records hold and what the methods section states. An FDA Global Substance Registration System search for FOXO4 returned nothing, so no UNII exists to check the CAS against.

The CAS number itself deserves a qualifier. It appears on both PubChem records, but in synonym lists deposited by chemical suppliers rather than in any regulatory or registry document, and it is the only identifier here with no independent second source. It is recorded above because a supplier-deposited synonym on a primary chemical database is still a checkable trail; it is flagged because a supplier-deposited synonym sits in a different class of evidence from a computed formula or an InChIKey.

What the 2017 experiments measured

Baar and colleagues arrived at FOXO4 by sequencing proliferating and irradiated human IMR90 fibroblasts and finding that senescent cells had raised the pro-apoptotic transcripts PUMA and BIM while lowering BCL-2. The reading was that senescent cells are primed to die and are held back by something else. FOXO4 was the candidate, and the peptide was designed to interfere with its interaction with p53 rather than to knock the protein down, because a short hairpin is not a therapeutic.

In culture, senescence was induced in IMR90 either by 10 Gy of X- or gamma-irradiation with analysis ten days later, or by two exposures to doxorubicin with analysis seven days later. The controls carry weight here. The same sequence in the L-isoform, FOXO4-L, and an unrelated D-retro-inverso peptide directed at FOXM1 were both run alongside, and cell density was tracked continuously in real time across the treatment window.

Three mouse settings followed: doxorubicin-exposed p16::3MR mice, fast-ageing XpdTTD/TTD mice crossed into the same reporter line, and naturally aged p16::3MR animals of 115 to 130 weeks. Administration was three injections on alternate days, intravenous in the chemotoxicity arm and intraperitoneal in the ageing arms. Group size was set by a power analysis at 80 percent power and alpha of 0.05, anchored on pilot measurements of plasma urea in 26-week wild-type versus XpdTTD/TTD mice, which returned seven animals per group.

The readouts were coarse and physical for the most part. Fur was graded before and after on a zero-to-four scale and expressed as a ratio to baseline; abdominal surface temperature by infrared thermometer stood in as a second measure of fur density, and only animals with a baseline above 34 degrees were included because the window was otherwise too small. Running wheel distance was scored at day 21, plasma urea and creatinine indexed renal filtration, and platelet count at sacrifice was compared to baseline. The abstract's tolerability statement is conditional in the original. Baar and colleagues write that the peptide neutralised doxorubicin-induced chemotoxicity "under conditions where it was well tolerated in vivo" - a clause that qualifies the result rather than reporting a tolerability measurement. No tolerability endpoint, dose-limiting observation or toxicity readout is reported for the peptide anywhere in the paper.

The 2025 structural work moved the target

Bourgeois and colleagues solved solution NMR structural models of the p53 transactivation domain bound to the FOXO4 forkhead domain and bound to FOXO4-DRI. Both partners in the second complex are disordered, and the paper describes a transiently folded assembly with no rigid interface. The peptide contacts the same residues on p53TAD2 that the forkhead domain contacts, and competes p53TAD2 away from it.

Isothermal titration calorimetry in that paper put the dissociation constant for FOXO4-DRI against p53 residues 1 to 94 at 400 plus or minus 280 nanomolar, and against the isolated TAD2 fragment at 18.5 plus or minus 3.6 micromolar. The FOXO4 forkhead domain bound p53 1-94 at 2.5 plus or minus 0.6 micromolar. The uncertainty on the tightest of those figures is seventy percent of the value, and the two FOXO4-DRI numbers differ by more than an order of magnitude depending on which p53 construct was titrated.

One finding cuts against the standard description of the molecule. The cationic segment is usually presented as a delivery tag bolted onto an active FOXO4 fragment, but residues within it contact p53TAD2, and they responded more strongly than the FOXO4-derived residues to phosphorylation of p53TAD2 at any of the sites tested, Ser46, Thr55 or both, while most other residues showed stronger chemical shift perturbations in the presence of unphosphorylated p53TAD2. Phosphorylation raised affinity for both the forkhead domain and the peptide. On this account the target is p53, not FOXO4, and the delivery segment is part of the binding surface.

None of this unsettles the 2017 cell-biology result, which was a selectivity observation and did not depend on knowing where the peptide sat. It does change what a shortened or re-engineered version can be expected to do. A construct that trims the cationic segment to reduce size or cost would be removing part of the interaction surface, not just a delivery handle, which is a different design problem from the one the field has generally described.

Where the follow-up work went

The testis was the first tissue taken up. Zhang and colleagues reported in 2020 that naturally aged male C57BL/6 mice of 20 to 24 months, six per group, given 5 mg/kg intraperitoneally on alternate days for three administrations, showed raised serum testosterone at thirty days with no difference in body or testis weight. Li and colleagues extended the same model in 2024 and reported changes in sperm quality and in secreted senescence-associated factors from Leydig cells.

Lung work attracted four separate groups. Han and colleagues in 2022 and Liu and colleagues in 2023 both used bleomycin-induced pulmonary fibrosis in mice; Meng and colleagues in 2021 reported effects on senescence-like cancer-associated fibroblasts and on radiation-induced fibrosis; and Jing and colleagues in 2024 gave the peptide intraperitoneally to Sprague-Dawley rat pups on postnatal days 4, 6, 8 and 10 in a hyperoxic bronchopulmonary dysplasia model, reporting outcomes similar to the two comparator agents in the same study.

Other tissues followed the same pattern: short courses, disease models in place of ageing itself, and a single peptide concentration in culture. Huang and colleagues tested expanded human chondrocytes, Kong and colleagues tested keloid organ cultures and fibroblasts, and Hu and colleagues tested aortic and endothelial senescence in aged and progeroid mice. A separate group designed a different peptide, ES2, by molecular modelling of the FOXO4-TP53 interface (Le 2021, EBioMedicine, PMID 34689087). The three-to-seven-fold figure originates in that paper as a cell-culture comparison: Le and colleagues report that their designed peptides showed three to seven times higher senolytic activity than FOXO4-DRI in doxorubicin-senescent A375 melanoma cells at 8 micromolar, in a single figure panel. A 2021 commentary on that paper (Tripathi 2021, PMID 34768086) restated the same range as applying to ES2 specifically and to both in vitro and in vivo work, which is wider than the primary figure panel supports; the commentary carries no EC50 values and no abstract in PubMed.

The findings that cut the other way

Born and colleagues published the sharpest counterweight in Circulation in 2023. Senescent pulmonary endothelial cells accounted for about thirty percent of lung senescent cells in normoxic mice, and clearing them made the measured outcome worse. Elimination by suicide gene or by ABT263 raised right ventricular systolic pressure and the hypertrophy index, increased vessel remodelling, and markedly decreased lung pulmonary endothelial cells. The peptide arm is reported more loosely: serotonin-transporter-overexpressing mice given either ABT263 or FOXO4-DRI showed pulmonary haemodynamic alterations and loss of pulmonary endothelial cells, with no per-arm pressure or hypertrophy figure given for FOXO4-DRI in the retrieved abstract. The accompanying editorial asked whether senolytics were guilty of overkill.

Huang and colleagues reported a null result inside a positive one. Treatment at 25 micromolar for five days removed more than half the cells in chondrocytes expanded to population doubling level 9 while leaving PDL3 chondrocytes largely intact, which is the selectivity the peptide was designed for. Chondrogenic potential did not improve. Cartilage generated from pre-treated cells carried lower SOX9 and COL2 expression and higher p16, p21 and p53 protein than cartilage from untreated cells.

Three record-keeping defects belong in the file. Han and colleagues 2022 carries a 2024 correction: the FACS scatter panels for one treatment group in Figure 6B had been duplicated from another, and the authors state the scientific conclusion is unchanged. Zhang and colleagues 2020 gives its in vitro concentration as 25 mM twice, in the text and the figure legend, where every comparable study used 25 micromolar. Huang and colleagues 2021 states that chondrocytes came from eight donors and then lists six ages.

None of the three is a retraction and none is an expression of concern. PubMed publication types and correction links were checked on every paper cited on this page and returned one erratum and nothing else. The unit discrepancy and the donor count are the kind of defect that survives peer review because neither changes a conclusion, and both are the kind that a reader reconstructing an experiment from the paper would run straight into.

The human record

The human record is empty. A ClinicalTrials.gov term search for FOXO4 returns a single unrelated lymphoma study of celecoxib, NCT07494565; an intervention search for FOXO4, and both intervention and term searches for FOXO4-DRI, return zero records, checked 18 August 2026. No registered study has administered this peptide to a person, at any dose, for any indication. Nothing has been registered on pharmacokinetics, tolerability or dose-ranging, and no observational study measures it in circulation, which is a further gap: unlike some compounds in this category, there is no human biomarker literature to fall back on because the molecule is synthetic and does not occur in people.

Human material enters the literature only as tissue. Chondrocytes were taken from donors and expanded in culture. Fourteen testis specimens, six from men aged 22 to 30 and eight from men aged 66 to 87, were examined by Zhang and colleagues for FOXO4 localisation, not treated with anything. Keloid tissue, autopsy lungs from infants with bronchopulmonary dysplasia, and lung protein from patients with pulmonary arterial hypertension were all characterised; none was dosed.

The regulatory position follows from that. FOXO4-DRI is not an approved medicine in any jurisdiction, has no marketing authorisation anywhere, and carries no FDA substance registration record. A 2026 review discussing the FOXO4-p53 axis as a target in brain ageing supports its human-relevant passages with trials of high-dose fisetin, a flavonoid taken by mouth with a different mechanism and a different evidence base. Findings for one senolytic do not transfer to another, and findings for a compound with registered human trials do not transfer to a compound with none.

Nine years separate the 2017 paper from this record, and in that time the compound has not entered a registered trial while several other senolytic classes have. The literature has instead widened sideways, into pulmonary fibrosis, keloid, testis, cartilage, aortic endothelium and neonatal lung, largely in short courses in rodents by groups with no connection to the originating laboratory. It remains a preclinical tool compound.

What is not known

No human has received this peptide in a registered study. A ClinicalTrials.gov term search for FOXO4 returns one unrelated lymphoma trial and an intervention search returns none, and searches on FOXO4-DRI by either field return nothing, so there is no pharmacokinetic, tolerability or dose-ranging data in people of any kind. There is also no published pharmacokinetic work in animals: no half-life, no clearance, no tissue distribution and no protease-stability assay for this molecule could be located, which leaves the central design claim of the retro-inverso approach uncharacterised for this sequence. Chronic administration has not been studied; every rodent regimen located ran to three or four injections with readouts at 21 or 30 days. Reproductive and developmental toxicity, carcinogenicity, immunogenicity against a foreign D-peptide, and behaviour in any non-rodent species are all absent from the record. The selectivity that the whole proposition rests on has been demonstrated against a small set of senescence models in a small set of cell types, and Born 2023 shows that clearing senescent cells is not uniformly beneficial: in the pulmonary vasculature it made the measured outcome worse. Whether the 400 nanomolar affinity reported by Bourgeois 2025 or the 18.5 micromolar figure from the same experiment is the relevant one in a cell has not been established.

Questions