Muscle-growth booster peptide (DF-3)
A short experimental peptide carved from follistatin that blocks myostatin, the body's natural brake on muscle growth, boosting muscle in lab studies. Experimental, not yet an approved drug.
A researcher, an agent, or an algorithm wrote down the sequence and picked a target to hit.
An AI model like OpenFold3 or AlphaFold built a 3D structure and scored how well it fits the binding site.
A second contributor repeated the computation on their own hardware and the scores matched.
A chemistry service or a researcher ordered the sequence, it was manufactured, and mass spectrometry confirmed the right molecule was produced.
A binding or activity measurement confirmed that it actually does what the computer predicted — or didn't.
What this is
DF-3 is a short (14 amino acid) experimental peptide that blocks myostatin, the body's own brake on muscle growth. It was carved out of a larger natural protein called follistatin — specifically, the loop within follistatin that grabs onto myostatin to neutralise it. Researchers stripped that loop down to its minimum functional core and found that a 14-residue fragment, capped with an amide on its tail, still inhibits myostatin in cells and increases skeletal muscle weight in mice (Saitoh 2020). The stored sequence VNDNTLFKWMIFNG corresponds to follistatin residues 41–54; the active peptide additionally carries a C-terminal amide (-NH₂) that protects the C-terminus from carboxypeptidases and is essential to the published activity. DF-3 is a research compound — it has not been tested in humans.
History
Myostatin (also called GDF-8) is a TGF-β-family growth factor that acts as a negative regulator of skeletal-muscle mass, making it a long-standing target for muscle-wasting indications. Follistatin, a glycoprotein found in serum and tissues, is one of two endogenous myostatin inhibitors (the other being myostatin's own prodomain) and is known to bind and inhibit several TGF-β family members including myostatin and activin A (Saitoh 2021).
The DF-series peptides were developed by Saitoh and colleagues, who set out to find the shortest fragment of follistatin's N-terminal domain (ND) that retained myostatin-inhibitory activity. Walking the sequence inward from a 19-residue parent (DF-1, follistatin 36–54), they found that the 14-mer (residues 41–54), named DF-3, was the minimum core — shorter peptides lost activity, and an N-terminal acidic stretch present in a longer variant (DF-2) actually reduced potency (Saitoh 2020). DF-3 was published as "the shortest peptidic myostatin inhibitor reported to date" (Saitoh 2021). A follow-up SAR study explored hydrophobic-residue substitutions to improve potency and to replace the oxidation-prone Met50 (Saitoh 2021). Independent work from Takayama and colleagues identified myostatin-inhibitory peptides from a different follistatin sub-domain (FSD1) over the same period (Takayama, Chem Pharm Bull 2020; Takayama, J Med Chem 2021), establishing follistatin's binding loops as a productive scaffold for short myostatin antagonists.
DF-3 is the parent of the DF-series lineage; later analogues on this platform are derived from it.
What it does
In cells, DF-3 blocks the signal that myostatin sends through its receptor, so that the downstream gene programme telling muscle cells to limit growth is dampened. In a luciferase reporter assay in HEK293 cells, DF-3 inhibited myostatin signalling in a dose-dependent way across 0.3–10 μM and was profiled in the same assay against activin A and TGF-β1, with the small-molecule TGF-β type-I-receptor inhibitor SB431542 as a positive control (Saitoh 2020).
In mice, a single intramuscular injection of 20 nmol DF-3 into the tibialis anterior of C57BL6/J animals (40 μL of 0.5 mM peptide in saline with 1% DMSO) increased the weight of the injected muscle compared with albumin control when measured 28 days later (Saitoh 2020). This is the central in-vivo result behind the peptide's classification as a functional myostatin inhibitor.
Mechanism
Follistatin neutralises myostatin by wrapping around the mature growth factor and burying the surfaces it would otherwise use to engage the activin type II receptors. The 41–54 stretch of follistatin sits on the binding face that contacts myostatin's hydrophobic residues, and three residues in particular — Phe47, Ile51 and Phe52 of the follistatin sequence — were identified from earlier X-ray analysis of the myostatin–follistatin complex as the contact hotspots; DF-3 preserves all three (Saitoh 2020). Alanine-scan derivatives of DF-3 (V41A through G54A) were synthesised to map which residues are load-bearing for inhibitory activity, and the 2021 SAR study extended this with hydrophobic substitutions designed to boost potency and to remove the oxidation-susceptible Met50 (Saitoh 2021).
Because follistatin is a broad TGF-β family inhibitor (Saitoh 2021), DF-3's activity profile against activin A and TGF-β1 was characterised in the original report alongside its myostatin activity, rather than being assumed away by design (Saitoh 2020).
Evidence
- Human: No human trials published.
- Animal: Single-dose intramuscular injection in C57BL6/J mice increased tibialis anterior weight at 28 days versus albumin control (Saitoh 2020).
- In vitro: Dose-dependent inhibition of myostatin signalling in a HEK293 luciferase reporter at 0.3–10 μM, with parallel profiling against activin A and TGF-β1 in the same assay (Saitoh 2020); alanine scan and hydrophobic-residue SAR (Saitoh 2021).
Known effects
- Myostatin inhibition — In vitro (luciferase reporter, HEK293) and preclinical (mouse intramuscular injection).
- Activity against activin A and TGF-β1 — Profiled alongside myostatin in the original HEK293 reporter assay (Saitoh 2020); selectivity numerics not pinned to a single half-maximal value in the source.
- Skeletal muscle weight gain — Demonstrated for the parent peptide in a single mouse study (Saitoh 2020); not replicated in independent labs as of the cited literature.
Regulatory status
- US / EU: Not approved for any indication. No active clinical trials registered for "DF-3" as a myostatin inhibitor.
- Research use only. DF-3 is a published research compound; it is not a marketed drug, not a compounded product, and not in any current regulatory pathway.
Related peptides
DF-3 is the parent of the DF-series lineage of follistatin-derived myostatin inhibitors. Other peptides discussed in the source literature target the same biology from different scaffolds:
- Follistatin FSD1-derived myostatin-inhibitory peptides (Takayama, Chem Pharm Bull 2020; Takayama, J Med Chem 2021) — a parallel short-peptide series derived from a different follistatin sub-domain.
- Myostatin prodomain-derived core peptides — a separate lineage that uses myostatin's own prodomain rather than follistatin as the scaffold; SAR work on this class has been reported to yield derivatives roughly eleven times more potent than the parent (Saitoh 2020).
Open questions
- DF-3 has been characterised in one mouse study and one SAR follow-up by the same group; no independent replication of the in-vivo muscle-weight gain has been published in the cited literature.
- Pharmacokinetics (serum half-life, biodistribution, route of administration beyond direct intramuscular injection) are not reported.
- No single IC50/Ki/EC50 value is pinned to DF-3 in the cited sources; activity is reported as a 0.3–10 μM dose response in a HEK293 reporter.
- Cross-reactivity with activin A and TGF-β1 has been profiled in one assay only; broader TGF-β family selectivity is not characterised.
- No disease-model efficacy data (muscular dystrophy, cachexia, sarcopenia) and no human data exist for DF-3 or its direct SAR derivatives.
- Myostatin inhibition has drawn renewed clinical interest as a way to preserve lean mass during GLP-1-receptor-agonist weight loss (Prado, Lancet Diabetes Endocrinol 2024); whether short peptidic inhibitors such as DF-3 have a role in that context is unexplored.
Research directions for this peptide, selected from the current sources — hypotheses you can explore and model. None of it is proven yet; tap any one to see the full thinking.
If this peptide blocks the same brake signal in heart cells that it blocks in muscle cells, could it stop dangerous scar tissue from forming after a heart attack?
If true, this peptide could help heart attack survivors avoid heart failure, the leading cause of hospitalization in older adults. It would give an existing muscle drug a powerful new use.
Might this mini-peptide block muscle-loss signals without interfering with other important body signals that the full protein disrupts?
If true, patients could build muscle without the risks to bone density, fertility, or immune function that broader inhibitors carry. This matters for long-term treatment of chronic muscle wasting.
Could this short peptide block myostatin using just two key building blocks, instead of the large surface its parent protein uses?
If true, scientists could design even smaller, cheaper muscle-building drugs that fit the same hidden pocket. This could help people with muscle-wasting diseases like muscular dystrophy or age-related frailty.
If researchers linked this tiny muscle builder to a popular weight-loss hormone, could the combined molecule do both jobs at once?
If true, patients would need only one injection instead of two separate drugs. This would lower costs, simplify treatment, and ensure both muscle protection and weight loss happen together.
Might the protective cap on the peptide end do more than block enzymes: could it also help the peptide fold into the right shape?
If true, chemists could design better caps or tougher building blocks at that spot, creating a longer-lasting drug that needs fewer injections. This would make treatment more convenient for patients.
▸full evidence table8 metrics
| metric | value | tool |
|---|---|---|
| ipSAE d0chn | 0.742 | AF2-Multimer · ColabDesign |
| interface dG | -23.12 kcal/mol | PyRosetta |
| decoy gap rotated | 23.12 kcal/mol | PyRosetta |
| hotspot occupancy | 11/13 | |
| decoy gap hotspot ko F7A | 11.47 kcal/mol | PyRosetta |
| off target delta gdf11 | 2.1 kcal/mol | AF2-Multimer |
| off target delta acva | 9.8 kcal/mol | AF2-Multimer |
| ipTM | undefined |
▸structural qualityopenfold3
| metric | value | note |
|---|---|---|
| gpde | 0.745 | global PDE — lower = better |
| disorder | 0.144 | fraction disordered |
| chain pair ipTM (A, B) | 0.675 | interface quality |
▸3-letter notation
▸recipeopenfold3-mlx 0.3.1
| parameter | value |
|---|---|
| model | openfold3-mlx 0.3.1 |
| weights | aedd8f3eb814e392… |
| hardware | apple_m4_base_16gb |
| mlx version | 0.31.1 |
| python | 3.14.3 |
| random seed | 42 |
| msa strategy | colabfold |
| runtime | 251s |
| predicted by | mlx@peptide |
| predicted at | 2026-04-14 |
▸ lineage 1 parent · 1 fork
▸citationbibtex
@peptide{pep00127,
sequence = {VNDNTLFKWMIFNG},
target = {gdf-8},
author = {peptidemodel},
year = {2026},
status = {reproduced}
}