Semax vs Selank: A Research Comparison

Semax and Selank are often mentioned together: both are short synthetic peptides, a large share of both literatures comes from the same network of Moscow laboratories ^[5], and both end in the same three amino acids. This page describes laboratory research material and published preclinical work only. It is not for diagnostic or therapeutic use.

Both molecules are linear heptapeptides, seven residues each ^{[1, 2]}. The first four positions differ (Met-Glu-His-Phe in Semax, Thr-Lys-Pro-Arg in Selank); positions five to seven are identical, Pro-Gly-Pro. Three of seven positions are shared, all at the C-terminal end.

Semax is built from a fragment of adrenocorticotropic hormone (ACTH). In the UniProt record for human pro-opiomelanocortin, ACTH occupies residues 138-176; its residues 4-7 read Met-Glu-His-Phe and residues 4-10 read Met-Glu-His-Phe-Arg-Trp-Gly ^[6]. Semax keeps the 4-7 stretch and carries Pro-Gly-Pro where ACTH has Arg-Trp-Gly; that comparison is our own reading of the sequence records ^{[1, 6]}, not a quoted sentence. The literature calls it both an analog of ACTH(4-10) ^[10] and ACTH(4-7)PGP ^[3], meaning an ACTH(4-7) fragment with a C-terminal Pro-Gly-Pro tripeptide ^[8], and its developers state the Pro-Gly-Pro tail was included to resist peptidases ^[3].

Selank is described by its originators as a synthetic analog of tuftsin, the Thr-Lys-Pro-Arg fragment of the human immunoglobulin G heavy chain, extended at the C-terminus with Pro-Gly-Pro for metabolic stability ^[4]. Tuftsin itself is the tetrapeptide Thr-Lys-Pro-Arg ^[7], so Selank is tuftsin plus a three-residue extension. Selank was designed at the Institute of Molecular Genetics of the Russian Academy of Sciences with the V.V. Zakusov Research Institute of Pharmacology ^[4].

In both molecules, then, Pro-Gly-Pro is an addition to a short natural fragment, and in both cases the originating authors give stability as the reason for the tail (resistance to peptidases for Semax ^[3], metabolic stability for Selank ^[4]); that is their stated design rationale, not an independently tested property. The similarity stops at the tail: the N-terminal portions come from different parents, and a result for one peptide should not be assumed to hold for the other.

Each item describes a model and a readout. All of it is animal, cell-culture or biochemical work.

• Rat hippocampus. A 2006 study from the Institute of Molecular Genetics measured BDNF protein, trkB phosphorylation and BDNF and trkB mRNA in rat hippocampus, with a conditioned-avoidance readout, and reported differences between Semax-exposed and control animals ^[10].
• Rat cerebral-ischemia models. Several papers from the same group used rat middle cerebral artery occlusion (MCAO), a surgical model of focal cerebral ischemia, and profiled gene expression afterwards: neurotrophin and Trk-receptor mRNA in cortex over 72 hours, compared with the free Pro-Gly-Pro tripeptide ^[8]; genome-wide transcripts, with immune-response and vascular-system genes highlighted ^[9]; and RNA-seq after transient occlusion, which identified 394 differentially expressed genes between Semax and saline groups ^[3].
• Rodent neurochemistry. A study involving the Zakusov Institute of Pharmacology measured striatal serotonin- and dopamine-related neurochemistry and locomotor activity in rodents, reporting a higher level of the serotonin metabolite 5-HIAA ^[11].
• Rat chronic-stress model. A 2024 study in male rats measured sucrose preference, body and adrenal weight, hippocampal BDNF and forced-swim behavior in animals exposed to Semax or to a second ACTH(4-10) analog that is a melanocortin-receptor agonist, and reported differences in several measures but none in forced-swim immobility ^[12].
• In-vitro copper chemistry. Two papers from a group in Catania, Italy, studied Semax without animals: copper(II) binding and copper-induced cytotoxicity in SH-SY5Y and RBE4 cell lines ^[13], and, in 2025, Cu(II)-amyloid-beta complexes and reactive-oxygen-species production in an SH-SY5Y cell model ^[14].
• Mouse spinal-cord-injury model. A 2025 paper from Wenzhou Medical University in China used a spinal-cord-injury model in female mice plus a PC12 cell neuroinflammation model. Its abstract attributes the identification of the mu-opioid receptor as a Semax target to RNA-seq, network pharmacology and docking, and also reports a USP18 knockdown experiment; the target assignment is a proposal from a single study, and we read only the abstract ^[15].

The same rule applies: each item is a model and a readout.

• Rat frontal cortex. A qPCR panel of 84 neurotransmission-related genes (77 analysed after seven were excluded for low expression) showed expression changes in 45 genes at 1 hour and 22 at 3 hours across the Selank and GABA groups combined; 29 of the 1-hour changes and 17 of the 3-hour changes followed Selank, including genes that also changed after GABA. Changes correlated positively with those after GABA at 1 hour (r = 0.86) and negatively at 3 hours (r = -0.39) ^[4]. The authors’ proposed mechanism is a hypothesis.
• Cell culture. In IMR-32 human neuroblastoma cells, Selank alone produced no change in mRNA across an 84-gene GABAergic panel, whereas the rat frontal-cortex panel above reported changes; the response differed when Selank was combined with GABA or with a reference compound ^[19]. The two systems differ in species, tissue and conditions, so the results are not directly comparable ^{[4, 19]}.
• Brain-membrane binding. Radioligand-binding experiments on brain membrane preparations, from the originating institute, reported that Selank altered the binding of radiolabeled GABA, which the authors read as allosteric modulation of GABA receptors ^[18]. This is an in-vitro, single-group hypothesis, and we found no independent replication ^[5].
• Mouse spleen. qPCR of four inflammation-related genes over time reported an approximately 3-fold fall in C3 mRNA within 30 minutes and changes in Casp1, Il2rg and Xcr1, with the Gly-Pro fragment giving mostly similar profiles ^[16]; an 84-gene inflammation panel reported changes in 34 genes ^[17].

Among the experimental papers cited here, only one, a 2001 in-vitro study, tests both peptides: Semax, Selank and their pentapeptide fragments inhibited enkephalin-degrading enzyme activity in human serum, while the tri-, tetra- and hexapeptide fragments tested did not ^[20]. A PubMed title-or-abstract search returns 11 records that mention both peptides ^[5]. Among the papers cited here, the two literatures otherwise share methods (gene-expression panels in rodent tissue) more than models.

The record on both peptides is sizable but narrow, and these limits apply to everything above.

• One network accounts for a large share. Counts are approximate and come from an editorial author-and-address match run on 6 October 2026, not a systematic review. Roughly 60 percent of PubMed records with Semax in the title (83 of 133) and about half of those with Selank in the title (30 of 59) list an Institute of Molecular Genetics or Kurchatov Institute address, or have an author from the originating group matched by name ^[5]. Affiliation data are missing for 43 of the Semax records and 21 of the Selank records, so the match rests on author names for those.
• Few records come from outside Russia. Of 133 Semax-titled records, six carry a clearly non-Russian first affiliation: four from Catania (Italy), one from Wenzhou (China) and one hypothesis article from Taipei (Taiwan) ^{[5, 13, 14, 15]}. Of 59 Selank-titled records, 38 carry affiliation data, and the only one with a clearly non-Russian first affiliation is a narrative article rather than a primary-data study ^[5]. The independence of the remaining groups cannot be judged from affiliations alone.
• Little independent replication. In the records we reviewed, we did not find a study that repeats the transcriptomic or GABA-binding findings in a different laboratory; the non-Russian primary papers address other questions, such as copper chemistry ^{[13, 14]} and a spinal-cord-injury model ^[15].
• Language. Among title-or-abstract hits, 87 of 208 Semax records and 29 of 68 Selank records are in Russian-language journals ^[5].
• Models used. Every primary study cited on this page is rodent, cell-culture or biochemical work, and some animal studies used one sex only: male rats ^[12] or female mice ^[15].
• Indirect readouts. The dominant measurements are gene-expression panels and behavioral tests, mostly from single laboratories. A change in mRNA at one time point does not by itself establish a mechanism, and mechanistic readings such as the allosteric-modulation proposal for Selank ^[18] are the authors’ hypotheses; the cell-culture study found no change in the panel’s mRNA levels with Selank alone ^[19].
• No dedicated review. By PubMed publication type, only one record with Semax or Selank in the title is typed as a review, a 2006 Russian-language paper, and none in English; publication types are assigned by PubMed and can miss narrative articles ^[5].
• Preclinical scope. There is no established human safety profile for either peptide. This page covers preclinical laboratory literature only.

The two peptides differ in composition, so they differ in mass: about 813.9 g/mol for Semax and 751.9 g/mol for Selank, a gap of roughly 62, with monoisotopic masses of 813.348 and 751.434 Da ^{[1, 2]}. A measured intact mass near 813.9 is consistent with Semax and one near 751.9 with Selank. Comparing measured mass with expected mass is one way identity is tested; HPLC vs LC-MS explains the method.

A single HPLC purity figure answers a different question: the share of detected peak area that belongs to the main peak under the stated method. It describes how much of the sample elutes as one peak, not which peptide that peak is, and a 99 percent figure would read identically on a Semax report and a Selank report. Identity is a separate question, which mass data address.

PubChem lists a second Semax record (CID 122178) that shares the free peptide’s formula and mass and carries the synonym “Semax acetate(80714-61-0 free base)” ^[21]. For Selank, a separate diacetate record (CID 155804769) lists a different formula, C37H65N11O13, and a mass of about 872.0 g/mol, under synonyms such as “TP-7 (diacetate)” ^[22], while the main Selank record lists “TP-7” among its synonyms ^[2]. Salt forms can also carry their own CAS numbers: tuftsin diacetate is registered as 72103-53-8 ^[23] and the free tetrapeptide as 9063-57-4 ^[7]. A mass or CAS number quoted without the form it refers to is therefore ambiguous. The COA guide explains how to read a certificate of analysis; How Peptides Are Made covers synthesis.

Each compound has a research summary and a certificate-of-analysis page: the Semax research summary and Semax COA, and the Selank research summary and Selank COA. More side-by-side pages are listed in the research comparisons index. This page describes laboratory research material only and is not for diagnostic or therapeutic use; see the research use policy.