MID-35 is real, well-characterised, and has never been given to a human

MID-35 is a genuine research peptide with genuine published pharmacology. It is also a compound whose entire in vivo record is mice, whose every published administration went directly into the muscle being measured, and whose amino-acid sequence has never appeared in a source you can read.

MID-35 is not vapour. It has a named originating laboratory, a documented eleven-year chemical lineage, a measured potency figure, a described mechanism, and five primary papers behind it. That already puts it ahead of most compounds circulating under a code name.

It also has no human data of any kind, no registered clinical trial anywhere in the world, no published pharmacokinetics, no published toxicology, no published molecular weight, and no publicly readable amino-acid sequence. Both of those paragraphs are true at once, and holding them together is the entire point of this page.

What MID-35 is

MID-35 is a synthetic peptide sixteen amino acids long. The name decodes as Myostatin Inhibitory D-peptide. The 35 is the compound's designation within its authors' own series of analogues — it is not a milligram amount and it is not the chain length, which is 16.

Structurally it is a retro-inverso peptide. The amino-acid sequence runs in reverse and is built from D-amino acids instead of the L-amino acids that make up ordinary proteins. A molecule assembled that way is difficult for proteases to recognise and cut, which is the reason the authors built it: their earlier L-peptide, MIPE-1686, worked but was fragile in the body. MID-35 is the stabilised D-form of that idea, and one derivative in the series was selected because it carried an arginine.

The lineage did not start from a drug. It started from a fragment of the mouse myostatin prodomain — the natural inhibitory region that keeps myostatin switched off in circulation. A 2015 paper in the Journal of Medicinal Chemistry identified the minimum active fragment of that prodomain; every compound since, MID-35 included, is a chemically shortened and stabilised descendant of it.

Where the evidence stops

Every in vivo experiment ever published on MID-35 was performed in mice. Not primarily in mice — only in mice.

We checked the trial registry ourselves rather than inferring it from the absence of trial papers. A ClinicalTrials.gov query for MID-35 returns nothing. A query for MIPE-1686, the L-peptide it was derived from, also returns nothing. Nothing in this chemical family has ever entered a registered human study.

The mouse work is real work. The cachexia study used male C57BL/6J mice, 8 to 12 weeks old, carrying Lewis lung carcinoma tumours. The 2026 pharmacology study covered young, adult and aged mice. These are competent experiments. They are experiments in mice.

Every published administration went into the muscle being measured

This is the detail that gets lost when a compound moves from a paper to a vial.

In every published in vivo experiment, MID-35 was injected directly into a single hind-limb muscle of a mouse — the gastrocnemius in the cachexia study, the tibialis anterior in the hypertrophy studies — and that same muscle is the one that was then weighed. One study delivered it through the skin over the target muscle using iontophoresis, a small applied current, which is the same local idea by another route.

There is no published subcutaneous data. No intravenous, no intraperitoneal, no oral. In any species. The compound has only ever been placed into the specific muscle the researchers wanted to grow, which is the design of a laboratory tool for interrogating a signalling pathway, not of a systemic drug.

How it works — and what it is not

Myostatin restrains skeletal-muscle growth. Blocking it can be approached from either end: block the receptor the signal lands on, or trap the signal before it arrives.

MID-35 traps the signal. It is a prodomain mimic — it binds the myostatin ligand itself and prevents that ligand from engaging its type I and type II membrane receptors, which stops the downstream Smad2/3 signal. In the published assays the ligand was pre-incubated with MID-35 before being added to cells, which is how you test a trap rather than a receptor blocker.

So MID-35 is not an ActRIIB antagonist, and it is worth being precise about that, because the two approaches carry different problems. The Cancer Science authors make the contrast themselves: soluble ActRIIB decoys also capture BMP9 and BMP10, which regulate vascular function, and that has produced vascular effects. MID-35 does not touch that receptor. That is a genuine mechanistic distinction — it is not a safety finding, and nobody has run the study that would turn it into one.

If the ligand-trap mechanism sounds familiar, it is the same family as follistatin-344, which we already cover. The two are not interchangeable, and the difference is in the next section.

MID-35 is not a selective myostatin inhibitor

Compounds in this class are routinely sold on selectivity. MID-35's own literature does not support that framing for MID-35.

In a luciferase reporter assay run in HepG2 cells — a human liver cell line — MID-35 suppressed signalling driven by TGF-β and GDF-11 as well as myostatin. It did not affect activin signalling. So the selectivity profile is genuinely narrower than a broad kinase inhibitor and genuinely broader than "a myostatin inhibitor": three ligands in, one ligand out.

The authors treat the extra breadth as a possible advantage in cancer cachexia rather than as off-target noise, and in that context the argument is reasonable. It is still a different molecule from the one implied by the phrase "selective myostatin inhibitor", and anyone repeating that phrase about MID-35 is describing something the source papers do not.

For contrast: follistatin traps myostatin and activin A. MID-35 spares activin and hits TGF-β and GDF-11. Same family of mechanism, different set of ligands captured.

What the mouse studies found — and what they did not

The measured potency is an IC50 of 0.19 µM in a cell-based reporter assay, roughly twenty times more active than the earlier peptide-2 at 4.1 µM. Sub-micromolar is respectable for a peptide of this size. It is not antibody-class affinity, and the figure comes from a reporter assay, not from an animal.

In the cancer-cachexia mouse model, injection into the gastrocnemius produced real muscle results: gastrocnemius weight relative to body weight rose significantly, atrophied muscle fibres improved, and grip strength rose significantly.

The same paper is unusually clear about what did not happen. In those mice, MID-35 did not affect survival, did not affect body-weight change, did not affect tumour growth, and did not affect heart weight. Muscle improved; the disease did not.

Two further numbers deserve to be read exactly as printed. The survival benefit that appears in that paper's title belongs to the combination of MID-35 with anamorelin, and it was reported at p = 0.052 by chi-square test — above the conventional 0.05 threshold, in a study whose own stated significance cut-off was 0.05. It should not be described as a demonstrated survival benefit.

And in that same combination arm, tumour volumes trended larger than in the control mice, at p = 0.09. The authors note that neither drug promoted tumour-cell growth in a dish, and they explicitly leave the question open for further evaluation. A trend at p = 0.09 in a small mouse study is not a finding. It is also the only oncology-adjacent observation in the entire MID-35 literature, and it points the wrong way. We would rather print it than leave a reader to discover it.

The 2026 study: slow onset, and an age effect

The most detailed pharmacology to date was published in June 2026, in tibialis anterior muscle of young, adult and aged mice.

The timing is the interesting part. Differentiation markers (Pax7, Myod1, Myog) rose and atrophy markers (Trim63, Fbxo32) fell within three days — but measurable muscle-weight gain did not appear until fourteen days later. Once it appeared, it was sustained for twelve weeks. Centralised nuclei and Pax7-positive signal indicate the growth ran through satellite-cell-mediated regeneration rather than direct fibre swelling.

The age result is the one that matters for anyone reading this as an anti-sarcopenia story. Sphingosine-1-phosphate, a lipid the authors link to satellite-cell activation, rose significantly on day 3 in young and adult mice. In aged mice it did not rise significantly. The population a muscle-preserving compound would be aimed at is the population in which part of the proposed mechanism did not fire.

Three things that have never been published

These absences are the practical content of this page.

1. The amino-acid sequence

The two papers that define MID-35 — the 2022 paper that names it and the 2026 paper that characterises it — are both in American Chemical Society journals, and we confirmed directly that neither is open access. The two open-access papers in the set describe MID-35 only as sixteen amino acids, arginine-containing and retro-inverso, and refer to the paywalled paper for its synthesis.

So there is no public sequence. If you are holding a vial labelled MID-35, there is no free source you can check it against, and neither can a laboratory.

2. A molecular weight

None of the papers prints one. This follows straight on from the first absence, and it is the one with teeth: a mass-spectrometry identity test works by comparing measured mass against the theoretical mass of the declared compound. With no published sequence and no published molecular weight, there is no reference figure for that comparison to be made against. A certificate can report a mass; nothing public establishes what that mass should be. Our guide to reading a COA covers why identity and purity are separate questions — for this compound the identity question currently has no public answer key.

3. Any pharmacokinetics

No half-life, in any species, by any route. No bioavailability, no clearance, no Cmax, no exposure data. The only stability work anywhere in this lineage is in vitro: a 2020 paper tested the L-peptide predecessor MIPE-1686 against isolated digestive enzymes in a tube. The 2022 abstract describes MID-35 as having considerable stability against biodegradation and attaches no figure to it in any source we can read.

We have refused to print a half-life for a compound before, for the same reason. A number that does not exist cannot be made to exist by rounding one that does.

Note what this section is not saying. No published toxicology is not a clean toxicology record — it is an absent one. The absence of harm data and the presence of safety data are opposite things.

A trap inside the cachexia paper

Worth flagging, because it is how MID-35 could acquire a human record it does not have.

The cancer-cachexia study is a two-drug paper. The other drug is anamorelin, an approved medicine in Japan with a real human clinical record, and the paper is full of human clinical language — plasma concentrations from human trials, references to randomised controlled trials. All of that belongs to anamorelin. None of it belongs to MID-35. Anyone summarising that paper quickly can transfer one drug's human evidence onto the other, and the result would read as though MID-35 had clinical support. It does not.

One laboratory, no independent replication

The full chain runs 2015 to 2026 and stays inside one collaboration: Kentaro Takayama and Yoshio Hayashi at Tokyo University of Pharmacy and Life Sciences, with Keisuke Hitachi and Kunihiro Tsuchida at Fujita Health University.

• 2015 — the minimum inhibitory peptides from the mouse myostatin prodomain (J Med Chem), and the 29-residue inhibitory core shown to block both receptor classes (PLoS One). Mice.
• 2019 — chain-shortened peptides improving grip strength (ACS Med Chem Lett). Mice.
• 2020 — enzymatic stability of the 16-mer MIPE-1686 (Chem Pharm Bull). In vitro.
• 2021 — a proposed binding mode for the 23-mer (Bioorg Med Chem). Computational.
• 2022 — MID-35 is created and named (ACS Med Chem Lett); MID-35 in cancer cachexia (Cancer Sci). Mice.
• 2023 — iontophoretic delivery (Pharmaceuticals); photooxygenation D-peptide variants (RSC Med Chem). Mice and in vitro.
• 2026 — sustained hypertrophy and sphingolipid metabolism, including aged mice (ACS Pharmacol Transl Sci). Mice.

A 2023 review in Cells places MID-35 among candidates under preclinical investigation and asked, at the time, for exactly the aged-mouse experiment the 2026 paper went on to run. No review describes clinical data, because there is none to describe. No group outside this collaboration has published a replication of MID-35's effects.

This is not an accusation. Single-source science is the normal early state of a compound. It is a reason the evidence is thinner than five papers makes it sound: five papers from one group is a smaller thing than five papers.

What we are not printing, and why

The published mouse experiments state the amounts used. We are not reproducing them here.

Those figures are microgram-scale amounts placed directly into a single mouse hind limb. There is no route by which they translate to a person: no species scaling factor applies without pharmacokinetic data, and there is none; no systemic administration has ever been studied, so there is nothing to scale toward; and the entire published design measures the muscle it injected. Printing the numbers would let them be converted, and any conversion would be arithmetic dressed as evidence.

The papers are linked below. Researchers who need the figures can read them in their original context, which is where they mean what they say.

The short version

MID-35 is a real, competently characterised 16-amino-acid retro-inverso D-peptide that traps myostatin and also blocks TGF-β and GDF-11. It grows mouse muscle when injected into that mouse's muscle, with an onset around two weeks and an effect lasting three months, and it did not improve survival, weight or tumour growth in the one disease model tested. It has never been given to a human, has never entered a registered trial, has no published sequence, no published molecular weight, and no pharmacokinetics in any species.

Compounds sold under research code names are not FDA-approved medicines, and nothing above is a recommendation to take any amount of anything. If you track compounds, our free app works from the tested milligrams on a certificate rather than the number on the label — for a compound with no public reference mass, that gap is the whole story.

Frequently asked

Sources

• Development of Myostatin Inhibitory d-Peptides to Enhance the Potency, Increasing Skeletal Muscle Mass in Mice (ACS Med Chem Lett) — the paper that names MID-35 — 2022
• Combination therapy with anamorelin and a myostatin inhibitor is advantageous for cancer cachexia in a mouse model (Cancer Science) — 2022
• Myostatin Inhibitory D-Peptides Induce Skeletal Muscle Hypertrophy along with Alteration of Bioactive Sphingolipid Metabolism (ACS Pharmacol Transl Sci) — 2026
• Increasing Skeletal Muscle Mass in Mice by Non-Invasive Intramuscular Delivery of Myostatin Inhibitory Peptide by Iontophoresis (Pharmaceuticals) — 2023
• Identification of the minimum peptide from mouse myostatin prodomain for human myostatin inhibition (J Med Chem) — 2015
• The Inhibitory Core of the Myostatin Prodomain: Its Interaction with Both Type I and II Membrane Receptors, and Potential to Treat Muscle Atrophy (PLoS One) — 2015
• Chain-Shortened Myostatin Inhibitory Peptides Improve Grip Strength in Mice (ACS Med Chem Lett) — 2019
• Enzymatic Stability of Myostatin Inhibitory 16-mer Peptides (Chem Pharm Bull) — 2020
• Age Is Just a Number: Progress and Obstacles in the Discovery of New Candidate Drugs for Sarcopenia (Cells) — independent review classifying MID-35 as preclinical — 2023
• ClinicalTrials.gov search for MID-35 — no registered studies

For research peptide users tracking their own protocol. Not medical advice. Peptides referenced here are research chemicals, not FDA-approved drugs. Consult a qualified clinician for medical decisions.