P21 improved memory in normal, aged and Alzheimer's-model rodents
P21 improved memory and neurogenesis in four rodent studies. No human trial has tested P21, and no safe or effective human dose has been established.

By Jay Spall, chemist and biochemist
Disclosure: Jay is a co-founder of The Peptide App. This review discusses the studies cited below; it is not a comprehensive live trial registry or treatment recommendation. Development and regulatory status can change. The app’s tools organize records and arithmetic and do not validate a research product.

On this page
Key facts
| Question | Direct answer |
|---|---|
| What is P21, structurally? | A short synthetic peptide (Ac-DGGLAG-NH2) built from the active region of ciliary neurotrophic factor (CNTF residues 148 to 151), with an adamantane group added to the C-terminus to help it cross the blood-brain barrier [1]. |
| Has P21 been tested in humans? | No. Every study of P21 is a rodent study or a review of rodent studies. No published human trial, Phase 1 safety data or established human dose exists as of mid-2026 [5]. |
| What does the P21 animal data show? | In normal mice and rats, and in a transgenic Alzheimer's-model mouse, P21 was associated with improved learning and spatial memory, increased hippocampal neurogenesis, elevated BDNF and reduced tau hyperphosphorylation [1][2][3][4]. |
| What is P21's evidence grade, and why? | E, minimal. The mechanism is rational and the rodent data are consistent across four studies, and none of it involves a healthy human brain, human dosing or human safety monitoring [1][2][3][4][5]. |
| Is there a safe or effective human dose of P21? | No. No human pharmacokinetic data exist, so any dose used outside a lab is a guess rather than a figure derived from a study. |
| What is the biggest practical risk of using P21? | Sourcing, not biology. A vial bought as "P21" from a research-chemical seller has no verified identity, purity or sterility standard to check it against. |
5 sources cited. View sources
What is P21?
P21 is a short synthetic peptide, Ac-DGGLAG-NH2, built from residues 148 to 151 of ciliary neurotrophic factor, the fragment that carries the relevant biological activity [1]. An adamantane group is attached to its C-terminus.
CNTF itself is a full-length neurotrophic protein studied for effects on neuron survival and metabolism. Large proteins like CNTF face two chronic problems in drug development. They do not cross the blood-brain barrier efficiently when given peripherally, and giving the whole protein systemically invites off-target effects tied to its full receptor-binding surface.
The rational-design answer, applied across several neurotrophin families, is to isolate the small peptide loop responsible for the activity and strip away the rest of the protein. P21 is that answer applied to CNTF. Dihexa's evidence covers another engineered neurotrophic peptide.
Why does P21 carry an adamantane group?
The adamantane group is there to get a short peptide into the brain. Short peptides are often too polar and too quickly degraded to reach the brain in useful amounts, so the designers attached adamantane, a rigid, lipophilic cage structure, to the C-terminus.
Adamantane conjugation is a known strategy for boosting membrane permeability and metabolic stability in peptide design. It is the specific engineering choice behind P21's claims to CNS penetration and oral bioavailability, properties confirmed across the compound's preclinical literature [5].
Reducing a large neurotrophic protein to its functional core and re-engineering it to solve a druggability problem is interesting medicinal chemistry. It is also the entire scientific interest of P21 so far. On its own it is not evidence that P21 does anything useful in a healthy human brain.
Has P21 been tested in humans?
No. All five published sources on P21 are preclinical, and no clinical program, registered trial or human case series exists to weigh against them. A 2022 review by the same laboratory group that produced most of the rodent work states plainly that all evidence to date is preclinical, with no human trials initiated [5].
That review also confirms P21's blood-brain-barrier permeability and oral bioavailability [5]. Those are delivery properties, measured preclinically, not outcomes in people.
What does the P21 rodent data show?
Four rodent studies report improved memory and neurogenesis with P21, across normal, aged and Alzheimer's-model animals. The founding study gave P21 peripherally to normal adult mice and found improved learning, short-term memory and spatial reference memory, along with increased neurogenesis and maturation of new neurons in the hippocampal dentate gyrus [1]. That established basic activity and a target organ, the hippocampus, in a non-diseased animal.
A 2014 study moved into aging biology. Chronic oral P021 given by gavage for 88 days to 22- to 24-month-old female Fisher rats reduced age-related decline in spatial memory, restored hippocampal neurogenesis, raised BDNF expression and rescued synaptic deficits in cortex and hippocampus [2]. That is the closest the literature comes to an aging-brain model, and it remains a rat, a specific inbred strain and a specific 88-day window.
Two further studies used the 3xTg-AD transgenic mouse, bred to develop amyloid and tau pathology. Twelve months of dietary P021 starting at 9 to 10 months of age reduced tau hyperphosphorylation at major neurofibrillary pathology sites, lowered soluble amyloid-beta, and was associated with rescued cognitive performance, neurogenesis and synaptic plasticity measures in those mice [3].
A follow-up study began dosing earlier, at 3 months of age, before overt pathology developed, and continued to 21 months. It found rescued dendritic and synaptic deficits, increased neurogenesis and reversed cognitive impairment, measured at 9 and 15 to 16 months into treatment [4].
Four consistent rodent studies from related groups is a real signal for further research. It is not four data points that generalize to a healthy adult human brain, because none of them tested one. A TREK-1 blocker marketed as a BDNF mimetic shows how quickly a neurotrophic mechanism claim can outrun its evidence.
What did the P21 studies never test?
No P21 study tested a cognitively normal adult being pushed toward peak performance. Every efficacy result comes from an aged, cognitively declining or transgenic-disease rodent. The 2010 study is the partial exception, using normal adult mice [1], and even that is a mouse hippocampus after peripheral dosing, not a person trying to think faster at work.
None of the five sources reports a controlled, blinded human cognitive outcome. None reports a human maximum tolerated dose, a human half-life or a human safety margin. None reports what happens with chronic, unsupervised self-administration in an adult whose CNTF and BDNF signaling is already intact. Semax's cognitive evidence shows what a thin human record looks like once trials do exist.
Where do P21 dosing figures come from?
No study establishes a human P21 dose. The rodent protocols used oral gavage and dietary administration over defined multi-week to twelve-month windows under laboratory conditions [2][3][4], and no published human-equivalent conversion exists for them.
There is no defensible way to translate how much a rat received into how much a person should take. A forum thread recommending a dose in micrograms is not translating a study, because the underlying studies never established a human dose to translate.
What is the practical risk of buying P21?
The identity of the material is the risk. P21 has no approved status and no clinical pharmacopeia entry, so no independent assay standard exists for a buyer to verify that a vial labeled "P21" contains adamantylated Ac-DGGLAG-NH2 at any specific purity, rather than a related peptide, a degraded fragment or an inert filler.
Dosing an unverified compound at a self-selected amount is not replicating the rodent research. It is an uncontrolled experiment in one person, in which even the identity of the substance is an assumption. Reading a peptide certificate of analysis shows what such a document can and cannot establish.
What is still unknown about P21?
The unknown list is short and blunt:
- Pharmacokinetics and dose. Human pharmacokinetics, a human dose and a human safety margin are all unestablished.
- Interaction risk. Interaction risk has not been studied.
- Chronic neurotrophic signaling. The long-term effects of chronically modulating a neurotrophic pathway in an otherwise healthy brain are unknown.
- Side effects. The reported side effects in circulation, injection-site reactions, headache and overstimulation, are anecdotal rather than findings from a completed human safety trial.
Sources
Last updated

Written by
Chemist and biochemist. Co-founder and author, The Peptide App.
Jay is a chemist, biochemist and entrepreneur whose work connects scientific research with consumer health products. He has held Chief Science Officer and product development leadership roles and previously served as Chief Revenue Officer at Minicircle.
Keep reading

Compound evidence
PE-22-28 blocks TREK-1 and showed antidepressant-like effects in mice
PE-22-28 blocks the TREK-1 potassium channel and does not mimic BDNF. All of its evidence is preclinical, and no human trial or safety data exist.

Compound evidence
Semax sped neurological recovery in open-label Russian stroke trials
Semax sped neurological recovery in open-label Russian stroke trials. Its one healthy-adult study enrolled 20 men with no placebo arm or blinding.

Compound evidence
Dihexa eased memory deficits in rats and in Alzheimer's model mice
Oral dihexa improved memory in rat and mouse models but failed in a Huntington's rat model. It has no human trials, and its key mechanism paper was retracted.