Adamax

A name attached to two different molecules

A PubChem compound query on the name Adamax returns HTTP 404 and the message that no CID matches the name. The substance endpoint returns exactly one record, SID 513899171, deposited by a chemical supplier, carrying the synonyms Adamax and Ac-Met-Glu-His-Phe-Pro-Gly-Pro-Ala-Gly-OH, a catalogue number in place of a registry number, and no link to any compound record. The FDA Global Substance Registration System returns nothing for the name. ChEMBL returns nothing either, although ChEMBL holds no record for the parent heptapeptide under any name or InChIKey, so that particular silence proves little. No CAS number, formula or molecular weight survived verification, which is why those fields above are blank.

The dispute sits in the last two letters of the shorthand. Product and aggregator pages write the molecule as Ac-MEHFPGPAG-NH2 or Ac-MEHFPGP-AG-NH2. Read one way, AG is alanine followed by glycine, and the molecule is an acetylated nonapeptide with no adamantane anywhere in it. Read the other way, AG is the adamantylglycine pseudo-residue borrowed from the P021 shorthand Ac-DGGL(A)G-NH2, where the terminal unit is not two amino acids at all but 3-amino-adamantane-1-carboxamide. The supplier deposit takes the first reading. The name, which is built on the word adamantane, takes the second.

Arithmetic separates them cleanly. The acetylated heptapeptide extended by alanine and glycine as free acid is C44H61N11O13S, average mass 984.1, and the corresponding amide is 983.1. The same chain condensed instead with 3-amino-adamantane-1-carboxamide is C50H69N11O11S, average mass 1032.2. Both figures are in circulation: one aggregator publishes about 984 daltons and states plainly that this is the backbone without the cage, another publishes 1032.24 as the molecular weight without qualification. Run on P021, whose structure is deposited, the same method reproduces PubChem's 578.7 for CID 56599151, so the method is sound and the gap of roughly 48 daltons is a gap between two molecules rather than two conventions.

The literature search returns an optimisation algorithm

A PubMed search on Adamax returns 39 records. Every one is a machine-learning paper, because Adamax is also the name of a variant of the Adam gradient-descent optimiser, and the titles run from potato blight detection to network intrusion detection to brain tumour segmentation. A Europe PMC search combining the name with peptide, Semax or nootropic returns 18 records, all from the same computational literature. Combining the name with Semax in Europe PMC returns zero. A PubMed search for adamantane together with Semax returns zero, as does a search on the string Ac-MEHFPGP.

Registry searches are equally empty. A ClinicalTrials.gov query on Adamax returns no studies by general term and none by intervention. The same query on Semax returns nothing either, and that result needs care. The parent has been studied in people for three decades, largely in Russian-language journals, so an empty result in that registry records a registration practice rather than an absence of clinical work. The openFDA label and Drugs@FDA endpoints return no match for either name.

Nothing here is a subtle gap. For the parent heptapeptide PubMed returns 231 records, spanning rodent pharmacology, gene-expression work and Russian clinical reports. For the molecule sold under the modified name the count is not small, it is zero. The searches that produced that zero are written out in full in the untraced-claims section below, with the databases and the exact terms, so that anyone can rerun them and get a different answer if one exists.

One formal document names the adamantylated form

European patent application EP4404912A1, filed 26 July 2022 with a priority date of 22 September 2021 and published 31 July 2024, is assigned to Ferring BV, with Vlasenko and Merkulova named as inventors. Its family members are WO2023046336A1, US20250127843A1 and CN117979954A. The claims cover an aqueous pharmaceutical composition of the heptapeptide Met-Glu-His-Phe-Pro-Gly-Pro with an antioxidant, an osmolarity-regulating agent, an optional preservative and water, at pH 4.0 to 7.0, intended for nasal delivery. The description defines a pharmaceutically acceptable derivative to include an N-terminal acetyl group with either a C-terminal amidate or a C-terminal adamantane moiety, and lists the acetylated heptapeptide bearing a C-terminal adamantane by name.

That is the whole of it. The five worked examples all test the unmodified heptapeptide as a formulation: methionine or no methionine, pH 4.5 against 5.0 against 5.5, sorbitol or no sorbitol, EDTA or no EDTA, each stored fourteen days at 60 degrees and read out as change in impurity content by ultra-performance liquid chromatography. No experiment on the adamantane derivative appears anywhere in the document, in any species or any assay. The string naming that derivative is itself corrupted in the published text, rendered two different ways at two points, apparently an optical-character-recognition artefact of a terminal carboxamide.

Dates cut against the origin story that circulates. The compound is said to have been created by a nootropics vendor that closed in 2018; the priority date here is September 2021, well after that, so the patent is not the source of the molecule and no earlier document naming it was located. The application was in any case deemed withdrawn with effect from 29 November 2025. What it establishes is narrow but real: a pharmaceutical applicant wrote this structure down as a permissible derivative within a nasal formulation, and tested the parent instead.

What the parent was measured to do

Everything below concerns Semax, the unmodified heptapeptide, and is included because it is the entire basis on which claims about the modified molecule are made. Shevchenko and colleagues 2006 prepared the peptide tritiated at the C-terminal proline at 56 Ci/mmol and gave it intranasally at 50 micrograms per kilogram in 20 microlitres to non-bred white rats of 200 to 250 grams. Two minutes after administration, 0.093 percent of the introduced radioactivity per gram was recovered from brain, of which 80 percent was intact peptide and the remainder metabolites. Degradation was rapid and the tripeptide Pro-Gly-Pro predominated in the biological samples.

Neurotrophin readouts come from one laboratory and are region-specific rather than global. Dolotov and colleagues 2006, in Brain Research, reported a maximal 1.4-fold rise in hippocampal BDNF protein and a 1.6-fold rise in trkB tyrosine phosphorylation after a single 50 micrograms per kilogram application in rats, with exon III BDNF mRNA up threefold and trkB mRNA twofold, alongside more conditioned avoidance reactions. The same group, in the Journal of Neurochemistry that year, reported specific, calcium-dependent binding of the tritiated peptide to rat basal forebrain membranes with a dissociation constant of 2.4 plus or minus 1.0 nanomolar and a Bmax of 33.5 plus or minus 7.9 femtomoles per milligram of protein. Agapova and colleagues 2007 found the direction of change was not uniform: after intranasal dosing in male Wistar rats, hippocampal BDNF and NGF expression rose, brainstem and cerebellar BDNF rose, and NGF fell in frontal cortex.

Human work exists and is mixed. Gusev and colleagues 2018 followed 110 patients after ischaemic stroke, 43 men and 67 women of mean age 58.0 plus or minus 9.7 years, in early and late rehabilitation groups each split into treated and untreated subgroups, and reported higher plasma BDNF and faster Barthel index recovery in the treated group. Serdiuk and colleagues 2007 gave the peptide intranasally to 27 patients with motor neuron disease and reported that it did not influence the course of chronic partial denervation on needle electromyography or the clinical rating scales, while the total quality-of-life score improved with a maximum at day 10. PubMed types that second study as a randomised controlled trial while its own abstract describes an open-label trial in sequential groups. Lebedeva and colleagues 2018 imaged 24 healthy volunteers, 14 given the peptide and 10 placebo, and reported a larger rostral subcomponent of the resting-state default mode network in the treated group.

None of these studies administered an acetylated form, and none administered an adamantylated one. Every figure above was measured on the free-amine heptapeptide with a free carboxyl terminus. Carrying any of them across to the modified molecule is an assumption about what two chemical modifications do to binding, clearance and brain entry. How that assumption has fared where it was actually tested is the subject of the next section.

Acetylation and the cage are not free parameters

Magri and colleagues 2016 compared the heptapeptide with its N-acetylated form directly. Acetylation changed the copper coordination chemistry, converting a CuN4 chromophore into a distorted CuN3O species with a weak apical interaction from the methionine sulphur, shifting the formal redox potential positive and leaving the complex reactive towards ascorbic acid where the free-amine parent was redox-stable. In SH-SY5Y neuroblastoma cells, the acetylated peptide did not protect against copper-induced toxicity, which the authors attributed to the loss of the free N-terminus. Zinc complexation and zinc influx were comparable between the two. Blocking the N-terminus is therefore a substantive change to the molecule and not a stability tweak.

Evidence on the adamantane unit points the same way. Muller and colleagues 2019 built five protein kinase A inhibitor peptides containing diamondoid amino acids and measured binding by microscale thermophoresis with NMR and crystallography alongside: affinity and binding pose were largely conserved, one variant gained 6.5-fold potency, and another lost affinity, which the authors attributed to lost hydrogen-bond contacts in the first solvation shell. Pramod and colleagues 2024 introduced adamantylglycine into a cell-penetrating peptide and found minimal influence on membrane transport, the point of the modification there being high-affinity capture by cucurbit[7]uril rather than improved permeation.

So the design rationale is real and the outcome is not predictable from it. Adamantylation raises lipophilicity and can slow exopeptidase attack, which is why the cage appears in the P021 series and in approved small molecules; it has also been measured leaving membrane transport unchanged and lowering binding affinity. Whether it helps a given peptide is an empirical question about that peptide.

One paper would speak directly to part of this. Shevchenko and colleagues 2013, in Doklady Biological Sciences, reported on the proteolytic stability of acetylated Semax in various biological media, and a companion paper from the same group examined analogues differing at the N-terminal residue. PubMed serves no abstract for either, and the publisher returned an authorisation gate rather than the text, so what they measured is recorded here as unread rather than summarised.

Registry and legal position

There is no approval for this molecule anywhere. It holds no marketing authorisation, no substance registration in the FDA system, no ClinicalTrials.gov entry and no investigational designation located this session. It is not an approved medicine in any jurisdiction, and the only official document in which the structure appears is a lapsed patent application that reports no experiment on it.

The parent occupies a different position, and the difference is often elided. Semax is described in the primary literature by the laboratory that developed it as a peptide used in the treatment of ischaemic stroke, and the secondary record states it has been registered in Russia as a 1 percent nasal preparation since 1994 and included on that country's list of vital and essential medicines. The Russian state register was not queried directly this session, so that registration is recorded here as asserted rather than verified. Neither the FDA nor the EMA has approved it; openFDA returns no label and no application for the name.

Approval of a parent does not extend to a derivative, and in this case there is no evidence that the derivative sold is the derivative named. The single supplier deposit in PubChem describes a molecule ending in alanine and glycine with no cage at all, while the name, the marketing and the patent language all describe one that carries it. Those are different substances, and nothing located this session says which is in the vial.

What is not known

Nothing is known about this molecule from experiment. There is no published study of any kind on it, so there is no pharmacokinetic profile, no receptor binding, no toxicology, no reproductive or chronic-exposure data, no immunogenicity work and no human exposure recorded anywhere in the literature or in any trial registry. It is not established which of the two circulating structures is actually supplied under the name, and the two differ by an entire adamantane cage; a purchaser and a supplier could each be describing a different compound in good faith. The parent's record does not close these gaps and has its own limits: the neurotrophin work comes overwhelmingly from one Moscow institute, sample sizes are absent from most retrieved abstracts, the human studies are small, largely open-label and published in Russian-language journals, and where PubMed's publication typing and a paper's own abstract disagree about whether a trial was randomised, that disagreement is unresolved here. No study has measured what N-terminal acetylation plus C-terminal adamantylation together do to this sequence, which is the one experiment that would make the parent's data relevant.

Questions