ACE-031 1mg Peptide

ACE-031 is not a peptide. It is a recombinant Fc-fusion glycoprotein: the extracellular, ligand-binding region of human activin receptor type IIB (ActRIIB, gene ACVR2B) joined to the Fc portion of a human IgG1 antibody. Each chain runs to 343 amino acids and carries three N-linked glycosylation sites, and two chains pair into a disulfide-linked homodimer of roughly 90,000 g/mol. That is more than thirty times the mass of a typical research peptide, and it is the fact that governs everything else about the molecule: how it is made, how it is stored, how long it circulates, and how a laboratory can tell whether a vial contains it.

The molecule was built by Acceleron Pharma and carries the international nonproprietary name ramatercept. The mouse-reactive version used in the early animal work appears in the literature as RAP-031 or ActRIIB.mFc. Because the construct is glycosylated and disulfide-bonded, it can only be produced in a mammalian expression system: the published preparation was expressed from a stable Chinese hamster ovary cell clone and purified over Protein A. Solid-phase peptide synthesis cannot make it at all.

Two clinical programmes were run and one of them was stopped. A Phase 1 single ascending-dose study in 48 healthy postmenopausal women produced measurable increases in lean mass and thigh muscle volume. The Phase 2 programme in ambulatory boys with Duchenne muscular dystrophy was terminated in 2013 after the second dosing regimen, following nosebleeds and telangiectasias that were traced to the trap picking up ligands it was never meant to bind. That termination is the single most important fact about this molecule, and almost every page selling it omits it. ACE-031 is supplied here as a reference material for in vitro laboratory research only.

ACE-031 is a ligand trap. It is neither an antibody against myostatin nor a receptor antagonist in the ordinary sense: it is a copy of the receptor's own catching surface, cut loose from the cell and released into the circulation. Myostatin (GDF-8), activin A, activin B and GDF-11 all signal by first docking onto the extracellular domain of ActRIIB, which then recruits a type I receptor and phosphorylates Smad2 and Smad3. A soluble copy of that same domain competes for the ligands in solution. What it binds never reaches the membrane, the Smad2/3 cascade is not initiated, and the standing brake those ligands hold on skeletal muscle is lifted.

The consequence in tissue is anabolic and it is not fibre-type selective. In C57BL/6 mice treated for 28 days, mean body weight ran 16% above vehicle controls, and wet weights rose 33% in soleus, 44% in plantaris, 46% in gastrocnemius and 26% in extensor digitorum longus. Mean fibre cross-sectional area in soleus rose 22% in type I fibres and 28% in type II, with no shift in the myosin heavy chain isoform distribution. That even-handedness distinguishes the trap from selective myostatin inhibition, which acts predominantly on type II fibres.

The Fc half of the molecule is not inert packaging. It dimerises the construct so that each molecule presents two binding arms, and it engages the neonatal Fc receptor, which recycles IgG out of the endosomal degradation pathway. The measured mean terminal half-life in the Phase 1 study was 10 to 15 days, with area under the curve and peak concentration rising linearly across the 0.02 to 3 mg/kg range. A protein that lingers for two weeks is what makes broad ligand sequestration a sustained systemic condition rather than a transient one.

Promiscuity is the design flaw, and it sits in the binding surface itself. The ActRIIB extracellular domain also has affinity for BMP9 and BMP10, two circulating cytokines that signal through ALK1 and endoglin to hold vascular endothelium in a quiescent state. Loss of that signal is the molecular basis of hereditary haemorrhagic telangiectasia, whose defining features are epistaxis and dilated superficial vessels. A trap that removes BMP9 and BMP10 from circulation reproduces the deficiency pharmacologically. That is the accepted explanation for what ended the Duchenne programme.

One further mechanistic point is easy to miss. Work with a dual anti-ActRIIA/ActRIIB antibody showed that blocking either type II receptor alone achieves only partial signal blockade and only a small increase in muscle mass, because myostatin and the activins bind both ActRIIA and ActRIIB with different affinities. A soluble ActRIIB domain sequesters ligands upstream of both receptors rather than blocking one of them, which is why the trap and a single-receptor antibody are not interchangeable tools.

A research peptide is a chain of a few dozen residues at most, assembled on a solid support, purified by reversed-phase HPLC, and characterised by a single mass number. ACE-031 is none of those things. It is a 343-residue polypeptide per chain, glycosylated at three positions, folded with the intrachain disulfide pattern of a receptor ectodomain, and assembled into a covalent dimer through the IgG1 hinge. The published preparation was expressed from a stable Chinese hamster ovary clone carrying a tissue plasminogen activator signal sequence, captured on Protein A, and dialysed into 10 mM Tris with 137 mM sodium chloride and 2.7 mM potassium chloride at pH 7.2.

Every one of those features has a laboratory consequence. Glycans mean the mass is heterogeneous, so a single number on a certificate is an approximation rather than a measurement. Disulfide-dependent folding means denaturation is a real failure mode and a denatured preparation still weighs the same. A covalent dimer means reducing conditions destroy the functional unit. And because activity lives in a folded surface rather than a sequence, purity and identity are separate questions: a preparation can be 99% one protein and still not be ACE-031.

This is also why the boilerplate that follows small peptides around does not transfer. Advice to hold lyophilised material at ambient temperature for shipping, or to keep a stock at room temperature, is written for a synthetic peptide and is not appropriate for a glycoprotein whose value is entirely in its conformation. Protein reference materials of this class are handled cold, protected from repeated freezing and thawing, and kept away from the air and liquid interfaces where Fc fusions aggregate.

Four registered studies exist. Two Phase 1 studies in healthy postmenopausal women, NCT00755638 with 48 participants and NCT00952887 with 70, both completed. Two Phase 2 studies in ambulatory boys with Duchenne muscular dystrophy, NCT01099761 with 24 participants and the extension NCT01239758 with 11, are both recorded as terminated, each with the same registry note that the decision was based on preliminary safety data.

In the published account, the Phase 2 study used an ascending-dose design with administration every two to four weeks and a primary objective of safety evaluation. The trial was stopped after the second dosing regimen. The reported reason was not efficacy failure and not a serious or severe adverse event: it was epistaxis and telangiectasias, described by the investigators as non-muscle-related findings. The pharmacodynamic signals were in the expected direction, with trends toward higher lean body mass, lower fat mass, higher bone mineral density and maintenance of six-minute walk distance, none of which reached significance in a cohort that small and that short.

The mechanistic explanation converges on BMP9 and BMP10. These two circulating ligands act on vascular endothelium through ALK1 and endoglin to maintain a quiescent, non-angiogenic state; loss-of-function mutations in ENG, ACVRL1 or SMAD4 produce hereditary haemorrhagic telangiectasia, in which the cardinal features are recurrent nosebleeds and dilated superficial vessels. A soluble receptor domain broad enough to trap myostatin and the activins is also broad enough to trap BMP9 and BMP10, and doing so phenocopies that disorder pharmacologically.

The strongest support for that account comes from a different molecule entirely. STM 434, an activin A ligand trap, was taken into a first-in-human Phase 1 study in 32 patients with advanced solid tumours. Mucocutaneous bleeding was among the most common treatment-related events, with epistaxis in 34% and gingival bleeding in 22%, and the investigators attributed it directly to off-target inhibition of BMP9. Two independently developed traps in two unrelated indications producing the same bleeding phenotype is what turns a single-trial safety signal into a property of the target class.

In 2025 an anti-doping laboratory published the first systematic chemical audit of ACE-031 sold as a research chemical. Fourteen different products were purchased from suppliers in the United Kingdom, continental Europe, China and the United States, and analysed by SDS-PAGE, Western blotting, IdeS protease digestion and high-resolution mass spectrometry. None of the fourteen contained ACE-031.

Twelve products carried an ACVR2B-immunoreactive protein with a main band near 58.4 kDa, close enough to a genuine ActRIIB-Fc fusion to pass a superficial gel comparison. Shotgun proteomics broke the resemblance: the tryptic peptides included sequences from the cytoplasmic region of ACVR2B, which a fusion built from the extracellular domain cannot contain, and the expected human IgG1 Fc peptides were absent. IdeS protease, which cleaves IgG below the hinge and releases the Fc fragment, could not cut the products at all. The observed mass matched the calculated 55.9 kDa of non-glycosylated full-length ACVR2B without its signal peptide, which points to bacterial rather than mammalian expression, and eleven of the twelve carried a His-tag left over from purification.

The remaining two products were something else again. One was identified by anti-follistatin immunoblot as follistatin-344. The other contained no protein at all: mass spectrometry found the growth hormone secretagogue ipamorelin. Across the whole set, the material being sold was full-length receptor, a different ligand-binding protein, or an unrelated peptide, and every one of the twelve receptor products carried many additional proteins alongside the main band.

The practical lesson is about what an analytical certificate can and cannot establish. A reversed-phase HPLC purity figure describes how much of one species is present, not which species it is, and for a glycoprotein it is close to meaningless. Establishing identity for this class needs SDS-PAGE under reducing and non-reducing conditions, an immunoblot against the extracellular domain, an IdeS cleavage test that a real Fc fusion must pass, and peptide mapping by mass spectrometry that reports Fc coverage and absence of cytoplasmic sequence. Any laboratory buying an Fc-fusion reference material should be asking for those, not for a purity percentage.

The platform did not die with the Duchenne programme. It survived by narrowing, and the successors are instructive because each of them fixes a different one of ACE-031's problems.

Luspatercept is the closest relative: another ACVR2B-Fc fusion, but with a modified extracellular domain that shifts the ligand set away from muscle and toward late-stage erythroid maturation. In a Phase 3 trial in transfusion-dependent beta-thalassemia, 21.4% of patients on luspatercept achieved at least a 33% reduction in transfusion burden during weeks 13 to 24 against 4.5% on placebo. Sotatercept took the other type II receptor, building the fusion from ActRIIA, and was tested in pulmonary arterial hypertension in a Phase 3 trial with 163 patients on active treatment and 160 on placebo, using six-minute walk distance at week 24 as the primary endpoint.

ACE-083 attacked the exposure problem instead of the ligand problem. It is a follistatin-288-Fc fusion, and the design exploits two properties of the follistatin-288 isoform: it binds heparin, so it stays where it is placed, and it is proteolysed quickly in circulation, so anything that escapes is cleared. In mice, intramuscular administration grew the targeted muscle without affecting the surrounding or contralateral muscle, in direct contrast to an ActRIIB-Fc fusion given by the same route, which acted systemically. Two randomised Phase 2 studies followed: in facioscapulohumeral muscular dystrophy the treatment difference in total muscle volume was 16.4% in the biceps brachii group and 9.5% in the tibialis anterior group, and in Charcot-Marie-Tooth disease type 1 it was 13.5% in the tibialis anterior. In both trials the muscle grew and the functional and patient-reported outcomes did not follow, and the programme was discontinued.

Read together, the successor record says two things about ACE-031. Broad systemic ligand trapping was the fastest way to add muscle mass and also the reason it could not be given safely, and adding muscle volume has repeatedly failed to translate into function in dystrophic and neuropathic muscle. Both are reasons to treat ACE-031 as a mechanistic probe rather than as an unfinished therapeutic.

Four strategies exist for interrupting the same axis, and they differ in how much of it they remove. A monoclonal antibody against myostatin neutralises one ligand. The myostatin propeptide sequesters that ligand by re-forming its latent complex. Follistatin binds myostatin plus the activins and GDF-11. A soluble ActRIIB domain sits furthest upstream of all of them: it captures whatever the receptor would have captured, which is the whole set plus BMP9 and BMP10.

That ordering predicts both the efficacy ranking and the safety ranking, and the safety ranking runs the other way. Selective myostatin antibodies have been clean and modest. The broad trap produced the largest mass changes in every model it was tested in and is the only one of the group whose lead clinical programme was stopped for a vascular finding.

There is also a species problem that explains why the mouse data oversold the human result. Circulating activin A levels in monkeys and humans are reported to be three to four times higher than in mice and rats. An agent that only removes myostatin is therefore working against a much larger residual ligand pool in a primate than in a rodent, which is one reason myostatin-selective agents that looked strong in mice underperformed in patients. It is also part of the argument for the trap: work with a dual anti-ActRIIA/ActRIIB antibody showed that blocking a single type II receptor gives only partial signal blockade and only a small increase in mass, and that complete neutralisation requires acting on both. Sequestering the ligand upstream achieves the same coverage, and inherits the same off-target exposure.

Blocking ActRIIB signalling changes more than muscle, and the non-muscle findings run in both directions. In mice on chow and high-fat diets, four weeks of RAP-031 increased lean and muscle mass, grip strength and contractile force, enhanced the ability of insulin to suppress hepatic glucose production in high-fat-fed animals, and raised adiponectin; ten weeks sharply reduced fat content on either diet. That result is the origin of most of the metabolic claims made about this molecule.

The counterexample is rarely quoted. In streptozotocin-treated mice, a model of insulin-dependent diabetes, ACVR2B:Fc increased body weight and muscle mass as expected but reproducibly worsened hyperglycaemia within about a week, alongside elevated serum corticosterone. Whether the trap helps or harms glucose handling depends on whether beta cells are present, which is a caution about generalising from diet-induced obesity models.

Bone and erythropoiesis move too. The Phase 1 study reported significant changes in serum biomarkers consistent with effects on bone and fat metabolism, and the Duchenne cohort showed a trend toward higher bone mineral density. Soluble ActRIIB decoy receptor increased hindlimb muscle weights and myofibre cross-sectional area in both the oim/oim and +/G610C mouse models of osteogenesis imperfecta, with improved contractile function in the oim/oim line. The erythroid arm of the same axis is the entire basis of luspatercept, which is why haemoglobin is a routinely monitored parameter in trap studies rather than an incidental one.

In neuromuscular models the pattern is consistent: mass responds, function often does not. In a severe spinal muscular atrophy mouse model, AAV-delivered soluble ActRIIB improved muscle mass and function, with a larger effect in fast fibre-type muscles than slow, but did not rescue the reduced motor unit number in tibialis anterior. Muscle size and motor unit integrity are separable, and this molecule addresses only the first.