Semax (MEHFPGP): A Research Guide to Its Identity, Binding Site and Preclinical Record

Semax is a synthetic heptapeptide with the sequence Met-Glu-His-Phe-Pro-Gly-Pro. Its first four residues reproduce the 4–7 segment of adrenocorticotropic hormone, and the C-terminal Pro-Gly-Pro tripeptide is not part of ACTH — which is why the peptide is often written in the literature as ACTH(4-7)PGP ^{[9, 11]}. The molecule is linear and unmodified — no disulfide, no cyclization, no lipidation — and carries a single methionine, a single histidine and a free N-terminal amine, three features that turn out to matter for how it behaves in solution and in binding assays.

Identity and characterization

The identity data below are the values carried in the verified identity map behind the Semax product page, where the molecular weight is recomputed from the formula on every build and a mismatch fails the deploy.

• CAS number: 80714-61-0
• Molecular formula: C37H51N9O10S
• Molecular weight: 813.9 g/mol
• PubChem CID: 9811102
• Sequence: Met-Glu-His-Phe-Pro-Gly-Pro (MEHFPGP)

The single sulfur in the formula belongs to the N-terminal methionine. It is worth noting because an earlier version of this catalog's identity data omitted it, which produced a molecular weight that did not match the structure — the kind of error that is invisible on a label and obvious to a mass check.

Mechanism as studied

Binding in membrane preparations

The most direct evidence that Semax interacts with a defined site comes from radioligand work on plasma membranes isolated from rat basal forebrain. Binding of tritium-labeled Semax was time-dependent, specific and reversible, required calcium ions, and gave a dissociation constant of 2.4 ± 1.0 nM with a BMAX of 33.5 ± 7.9 fmol/mg protein ^[1]. That is a nanomolar site with a low density, measured in a tissue homogenate.

What the experiment does not deliver is a molecular identity. The authors of that study concluded only that their results point to the presence of specific binding sites; no receptor protein has been cloned, expressed and shown to reproduce those constants, so "the Semax binding site" is defined by an assay rather than by a gene ^[1]. Semax is routinely grouped with the melanocortins because of its ACTH-derived core, and one study using it in parallel with a melanocortin receptor agonist describes Semax as a noncorticotropic ACTH(4-10) analog ^[10] — but no published affinity panel for Semax across cloned melanocortin receptor subtypes could be located in the searchable literature.

The peptide and its fragments act as a set

Semax does not stay intact in the presence of cells. Using uniformly tritium-labeled peptide and chromatographic separation of degradation products, the major products of MEHFPGP biodegradation in the presence of nerve cells were shown to be the pentapeptide HFPGP and the tripeptide PGP ^[5]. A companion study of the same system reported that the predominant processes in the presence of plasma membranes and of glial and neuronal cultures were the splitting away of ME and of GP and the formation of pentapeptides, and that the pattern of products differed between glial and neuronal cells ^[13]. Those fragments are not inert leftovers: they have their own binding sites on neuronal plasma membranes, and in competition experiments the five-residue fragments EHFPG and HFPGP showed the highest competitive activity at Semax sites, leading the authors to argue that these interactions are probably not limited to specific binding at orthosteric sites ^[4].

The practical consequence is that any assay run on Semax over more than a few minutes in a biological matrix is reporting the behavior of a mixture, not of one molecule. Studies that run Semax and its PGP fragment side by side exist for exactly this reason: in cultured cerebellar granule cells under glutamate exposure, Semax at 100 µM and PGP at 20 and 100 µM both delayed the development of calcium dysregulation and the reduction of mitochondrial potential, and incubation with these peptides improved neuronal survival by on average 30% ^[14].

Neurotrophin expression as the dominant readout

Most mechanistic work on Semax uses neurotrophin expression as its endpoint. In rat hippocampus, a single administration produced a maximal 1.4-fold increase in BDNF protein, a 1.6-fold increase in trkB tyrosine phosphorylation, and roughly 3-fold and 2-fold increases in exon III BDNF and trkB mRNA respectively ^[2]. In basal forebrain, BDNF levels rose three hours after administration, while cerebellum in the same animals did not change ^[1].

Those are modest, region-selective effects, and the direction is not uniform. A time-course study of NGF and BDNF transcripts in hippocampus, frontal cortex and retina found what the authors call multidirectional activation: expression of both genes fell in hippocampus and retina twenty minutes after administration while rising in frontal cortex, and in the retina NGF then stayed practically constant while BDNF was significantly increased at ninety minutes ^[3]. A summary of the form "Semax raises BDNF" flattens that into something the data do not say.

Cell-level effects

In primary cultures from rat basal forebrain, Semax may increase survival of cholinergic neurons approximately 1.5- to 1.7-fold in vitro, and it stimulated choline acetyltransferase activity at 100 nM, while the number of GABAergic neurons and the total count of neuron-specific-enolase-positive cells were unaffected; across 1 nM to 10 µM it did not change glial proliferation ^[6]. The hedge is the authors' own, and the selectivity within a single culture is the informative part of the result.

A separate electrophysiological study cultured dorsal root ganglion and dorsal horn neurons together for twelve days in the presence of Semax at 10 and 100 µM. Frequency of spontaneous glutamatergic postsynaptic currents rose, amplitude and frequency of miniature currents changed little, the paired-pulse ratio increased from 0.53 to roughly 0.91–0.95, and the coefficient-of-variation ratio fell — a pattern the authors read as increased multivesicular glutamate release ^[7]. Note the concentration: 10 µM is about 4,000 times the 2.4 nM dissociation constant in ^[1] and 100 µM about 42,000 times it, so the functional window sits roughly three and a half to four and a half orders of magnitude above the reported affinity, and nothing in the literature reconciles the two., and nothing in the literature reconciles the two.

Metal coordination

The histidine imidazole and the free N-terminal amine make Semax a competent copper ligand. At physiological pH the Cu(II)–Semax complex presents a CuN4 chromophore; acetylating the N-terminal amine shifts the main species to a distorted CuN3O chromophore with a weak apical interaction from the methionine sulfur, raises the formal redox potential, and makes the complex reactive toward ascorbic acid, whereas the free-amine complex was redox-stable and unreactive. Acetylation also did not protect against Cu(II)-induced toxicity in SH-SY5Y neuroblastoma cells, which the authors read as demonstrating the crucial role of the free N-terminus in that protection, while both forms bound Zn(II) with comparable strength ^[11]. Two things follow: the free N-terminus is not a spare part, and results obtained with N-modified analogs should not be transferred to Semax without checking.

Model systems it has actually been examined in

The published record is preclinical and concentrated in rodents.

• Membrane and cell-free preparations. Rat basal forebrain plasma membranes for radioligand binding ^[1] and for degradation work ^[13]; neuronal plasma membranes for the competition experiments on Semax and its fragments ^[4]; solution-phase characterization of the Cu(II) and Zn(II) complexes ^[11].
• Primary and immortalized cell culture. Rat basal forebrain neuron and glia cultures ^{[6, 13]}; DRG–dorsal horn co-cultures ^[7]; cerebellar granule cell cultures ^[14]; SH-SY5Y neuroblastoma cells ^[11]; a PC12 neuroinflammation model ^[12].
• Rodent occlusion models. A genome-wide transcriptional analysis in rat cortex after permanent middle cerebral artery occlusion found that Semax predominantly enhanced expression of immune-system genes; twenty-four hours after occlusion the genes implicated in that response represented more than half of the 68 genes whose expression Semax altered at that timepoint, whereas at 3 hours, when 96 genes were altered, the authors report only that several genes governing immune-cell activity were affected. Twenty-four vascular-system genes showed altered expression at 3 hours and 12 at 24 hours ^[8]. An RNA-seq study in the transient occlusion model identified 394 differentially expressed genes above a 1.5-fold cutoff at 24 hours relative to saline, with inflammation-associated transcripts suppressed and neurotransmission-associated transcripts activated — the opposite direction to the occlusion condition alone ^[9].
• Rodent stress paradigms. In a chronic unpredictable stress protocol in male Sprague-Dawley rats, repeated administration of Semax reversed or substantially attenuated stress-induced changes in sucrose preference, body-weight gain, adrenal mass and hippocampal BDNF; in the same study, no effect of either the stress procedure or the peptides on forced-swim immobility was detected ^[10].
• Mouse spinal cord injury. A 2025 study reported improved functional recovery and inhibited lysosomal-membrane-permeabilization-related pyroptosis in female C57BL/6 mice, and used RNA sequencing, network pharmacology and molecular docking to nominate the µ-opioid receptor as a Semax target regulating the ubiquitin-specific protease USP18 ^[12].

What the evidence does not establish

This section is longer than a promotional summary would make it, because the gaps are real.

• No identified receptor. A nanomolar, calcium-dependent binding site in a membrane homogenate ^[1] is not a receptor. Nothing in the literature names the protein, and the competition data suggest the interaction may not be a simple orthosteric event at all ^[4].
• Target nomination is not target validation. The µ-opioid receptor assignment in ^[12] comes from network pharmacology and molecular docking, both computational. No binding constant for Semax at that receptor is reported.
• A concentration gap nobody has closed. Functional effects in culture at 10–100 µM ^[7] sit far above the reported binding affinity ^[1].
• Differential expression is not mechanism. The transcriptome studies ^{[8, 9]} describe which transcripts move and in which direction. They do not establish which of those changes are causal, which are downstream, and which are noise at a 1.5-fold threshold.
• Narrow provenance. A large majority of the in vivo work carries overlapping authorship from one Moscow institute. That is not an accusation of error; it is a statement about how much independent replication exists, which is little.
• Single-sex designs. The stress work used male rats only ^[10]; the spinal cord injury work used female mice only ^[12]. Neither reports a sex comparison.
• Mixed endpoints within single studies. The same experiment that found changes in sucrose preference and hippocampal BDNF found none in forced-swim immobility ^[10]. Reporting only the positive arm would misrepresent it.
• No human or veterinary conclusions. Nothing in this guide supports use in people or animals. No dose, schedule or route of administration appears anywhere in this guide, and none should be inferred from the culture concentrations and timepoints quoted above, which are reported solely to identify the conditions under which a measurement was made.

Handling and stability

Whatever the intent behind the Pro-Gly-Pro extension, it does not make the molecule inert to peptidases. In the presence of nerve cells, the principal degradation products are HFPGP and PGP ^[5], and in the presence of plasma membranes and of glial and neuronal cultures the dominant events are loss of ME and of GP with formation of pentapeptides, in patterns that differ between the two cell types ^[13]. Semax is also not passive toward peptidases itself: in the tritium-label study it showed a high inhibitory effect on blood-plasma enkephalinases, attributed to action on aminopeptidases ^[5]. Any experiment involving plasma or cell lysate should treat that as a possible confound rather than a curiosity.

Chemically, the two reactive handles are the methionine sulfur, which is oxidizable, and the histidine and N-terminal amine pair, which coordinates Cu(II) readily enough that the geometry and redox behavior of the complex change when the amine is blocked ^[11]. Adventitious copper in water, buffer salts or glassware is therefore not a neutral background for this peptide.

Beyond that, the peer-reviewed record is thin on the things a laboratory actually wants: there is no widely cited dataset giving quantitative shelf life for the lyophilized powder, solution stability as a function of pH and temperature, or freeze–thaw tolerance for Semax specifically. The product page states that the material is supplied as lyophilized powder with refrigerated storage; that is a supplier specification, not a literature finding, and this guide does not dress it up as one.

Open questions

Four questions would change how this compound is interpreted if anyone answered them. What protein does tritiated Semax bind in basal forebrain membranes, and why does that binding need calcium ^[1]? Is the activity attributable to the heptapeptide, to HFPGP, to PGP, or to the sequence in which they appear ^{[4, 5, 14]}? Can the nanomolar affinity and the micromolar functional concentrations be reconciled within one mechanism, or are they two different phenomena sharing a name ^{[1, 7]}? And do the transcriptome-level changes reported after occlusion ^{[8, 9]} reflect a driving mechanism or a downstream signature of tissue that was damaged less?

Until those are addressed, Semax is best described as what it demonstrably is: a well-characterized small peptide with a measurable, calcium-dependent binding site of unknown molecular identity, a reproducible but modest effect on neurotrophin expression in specific rat brain regions, and a large preclinical transcriptomic literature produced largely by one research lineage.