Follistatin 344 vs 315: one protein, two numbers

What do 344 and 315 actually count?

Residues, not milligrams: both figures count amino acids in a protein chain. Neither is a concentration, a purity, or a version number. The UniProt entry for human follistatin lists the sequence at 344 residues and a computed mass of 38,007 daltons before any sugar is attached, and it annotates two asparagines as N-linked glycosylation sites.

The cut that produces 315: the same entry marks residues 1 to 29 as a signal peptide, the short address tag that carries a newly built protein into the secretory pathway and is then removed. Subtracting 29 from 344 leaves 315, which is exactly the span UniProt labels Follistatin.

Where the 344 label came from: cDNA sequencing. Shimasaki and colleagues screened a human testis library and reported that three of eight clones predicted a precursor of 344 amino acids while the remaining five predicted one of 317, the difference arising from alternative splicing of the precursor messenger RNA (Shimasaki, 1988). That paper also describes four contiguous domains encoded by precisely separated exons, three of them resembling each other and resembling epidermal growth factor.

Where the 315 label came from: protein chemistry, five years later. Sugino and colleagues purified six molecular forms of follistatin from porcine ovaries and described the core sequences as 315 and 288 amino acids, plus a 303-residue form that appears to come from FS-315 by cleavage of the twelve carboxyl-terminal residues (Sugino, 1993).

Why the two literatures use different numbers: one counted the transcript, the other counted the purified protein. A gene therapy construct is named for the sequence it carries, so it inherits the precursor number. A protein in a test tube is named for the chain you can weigh, so it carries the mature number. Nothing is being compared when 344 is set against 315.

If not 344 against 315, then what is the real split?

The splice: one gene, two messages, two precursors. The 344-residue precursor matures to FS-315. The 317-residue precursor matures to FS-288. Both mature chains carry the same core domains, and the arithmetic gap between them, 27 residues, sits at the carboxyl end of the longer one.

What that tail changes: Sugino's group reported that FS-288 bound heparan sulfate-Sepharose and FS-315 did not, and that COS cells transfected with FS-288 DNA held the protein on the cell surface while cells given FS-315 DNA secreted it into the medium, where it did not bind the surface at all (Sugino, 1993). One form stays where it is made. The other travels.

The affinity result people skip: in that same paper all six purified forms bound activin at roughly the same strength, a dissociation constant of 540 to 680 picomolar. What differed was potency in a cell system: suppressing follicle-stimulating hormone release from rat pituitary culture took an ED50 of 2 ng/mL for FS-288 against 20 ng/mL for FS-315, a tenfold gap between forms that bind the same ligand equally well.

Repeated thirteen years later: Sidis and colleagues built recombinant FST288, FST303, FST315 and follistatin-like-3 and found activin-binding affinities and kinetics comparable across the set, while cell-surface binding fell away sharply in the order FST288, FST303, FST315, then follistatin-like-3 (Sidis, 2006). In one bioassay the direction of effect reversed outright: FST288 suppressed activin-dependent cell proliferation and FST315 enhanced it.

And a third form the argument usually leaves out: Saito and colleagues describe three follistatin isoforms produced from the gene, FST288, FST303 and FST315, and reported FST315 secreted fastest with FST288 secreted more slowly and partly retained inside the cell (Saito, 2005). A 2016 review article describes FST315 as the predominant follistatin form, with the FST288 variant accounting for less than 5 percent of the encoded messenger RNA (Shi, 2016).

Is follistatin 344 stronger than follistatin 315?

The question has no answerable form, because the two names do not label two competing molecules. FST-344 is the precursor whose mature product is FS-315. Ranking one against the other is ranking a protein against the same protein one processing step later, and no study in the reference list below sets up that comparison, because there is nothing to set up.

The potency comparison that does exist: FS-288 against FS-315, and it does not run the way a strength ladder would suggest. Two independent groups thirteen years apart put activin-binding affinity at roughly the same level for both and located the functional gap somewhere else entirely, in how tightly each form holds to a cell surface (Sidis, 2006).

One conflict worth printing rather than smoothing: the 2015 review of the Becker trial states that the FS315 isoform has a tenfold lower affinity for activin than FS288 (Al-Zaidy, 2015). The two primary papers do not report that. Sugino measured dissociation constants inside a narrow band across six purified forms, and Sidis reported comparable binding affinities and kinetics across four proteins. We print the disagreement and leave it standing, because a review and the measurements underneath it are saying different things and we cannot resolve that from the outside.

What did the human follistatin trials actually use?

A gene, not a vial: every follistatin trial that has published a result delivered the FS344 coding sequence into muscle inside an adeno-associated virus. ClinicalTrials.gov record NCT01519349, a phase 1 study across Becker muscular dystrophy and sporadic inclusion body myositis, lists the intervention as rAAV1.CMV.huFollistatin344 and an actual enrollment of 15. That record posts no results and no participant flow, so how many of the fifteen were treated is not in the registry at all: the treated counts below come from its two published papers. A second record, NCT02354781, carries an actual enrollment of 3 Duchenne participants, posts a participant flow showing three started and three completed, and lists the intervention as rAAV1.CMV.huFollistin344, spelled that way in the registry entry. Both are marked completed.

A third completed record, and it is not a virus: NCT06411366 is a phase 1 open label single dose study in healthy adults whose intervention the registry describes as a follistatin plasmid, meaning a circular piece of DNA rather than a viral vector, named in the record as Follistatin-344 plasmid. Actual enrollment is 43, the largest enrollment of any completed follistatin gene transfer record returned by the registry search in reference 18, and it ran from August 2022 to August 2023. It posts neither results nor a participant flow, so how many of those 43 received the plasmid is not reported anywhere in the record, and it adds no outcome to compare against the AAV trials.

The Becker result: six patients received bilateral intramuscular quadriceps injections of AAV1.CMV.FS344, and the two cohorts were not followed for the same length of time, which matters because the doses also differed. The first cohort, at 3 x 10¹¹ vg/kg per leg and measured at 12 months, had two patients improve on the six-minute walk test by 58 meters and 125 meters, and a third by 9 meters, which the authors judged inside the range of variability for this population. The second cohort, at double that dose and measured at 6 months, had two improve by 108 meters and 29 meters while the third declined by 14 meters. The authors reported reduced endomysial fibrosis and more normal fiber size distribution on histology (Mendell, 2015).

The myositis result: six patients with sporadic inclusion body myositis received rAAV1.CMV.huFS344 into the quadriceps of both legs, alongside an exercise regimen. Annualized to a median one-year change, walk distance moved +56.0 meters per year against a decline of 25.8 meters per year in eight untreated subjects matched for age, gender and baseline measures (Mendell, 2017).

Two things that paper says twice, differently: the dose denominator and the p value. The abstract puts the dose at 6 x 10¹¹ vg/kg delivered to the quadriceps of both legs, while the discussion calls it a uniform 6 x 10¹¹ vg/kg per leg and the registry record lists its top cohort per quadriceps. Read per leg, the total vector delivered is twice what the abstract's wording implies, and nothing in the paper settles which reading is right. On the walk distance, the abstract and the results text both give p = 0.01 for that annualized comparison, and the annualized column of table 2 gives p = 0.001. We print both readings rather than pick one, because the source does not settle either.

The primate work underneath: before either trial, an AAV1-FS344 vector injected into the quadriceps of cynomolgus macaques produced increases in muscle size and strength that the authors described as pronounced and durable, with no abnormal changes in the morphology or function of the key organs they examined (Kota, 2009). That paper states its isoform choice openly: an alternatively spliced form of human follistatin that affects skeletal muscle while having only minimal effects on nonmuscle cells.

Why anyone is chasing follistatin at all: the muscle interest runs through myostatin, which is a different pathway from the growth hormone axis covered in our HGH comparison. Purified myostatin binds activin type II receptors, and that binding was inhibited by follistatin; transgenic mice expressing follistatin under a skeletal-muscle promoter showed muscle mass increases comparable to myostatin knockout mice (Lee and McPherron, 2001).

Why did the trials deliver a gene instead of injecting the protein?

Because the protein does not stay: a 2013 pharmacology paper set out to characterize follistatin's pharmacokinetic and pharmacodynamic behavior and concluded that the intrinsic properties of native FST315 are poorly suited to a parenterally administered biotherapeutic with broad systemic effects (Datta-Mannan, 2013).

What the same group had to build instead: they fused FST315 to a murine IgG1 Fc and removed its intrinsic heparan sulfate binding. The abstract summarizes the gain as roughly 100-fold longer terminal half-life and roughly 1600-fold greater exposure, and table 2 of the paper is where those two ratios come from: the single 3 mg/kg intravenous rows in mice, a terminal half-life of 1.2 hours for native FST315 against 104 hours for the engineered molecule, and an area under the curve of 569.4 against 924,963. The subcutaneous rows of the same table are narrower, 1.5 against 104 hours and 482.1 against 388,867. The weekly subcutaneous finding, a dose-dependent effect on muscle mass in mouse models of muscle atrophy and degeneration that native FST315 did not produce, is a separate pharmacodynamic experiment from those intravenous pharmacokinetics, not the same comparison read twice (Datta-Mannan, 2013).

What that means for a powder: a construct name is not a formulation. In the trial literature FS344 names a DNA sequence packaged inside a virus and expressed by muscle cells that then make and secrete the protein continuously. A powder is a different object from that, whichever number is printed on the label, and no source on this page characterizes what is inside one or what it does once it leaves the container. What the record does measure is a recombinant FST315 expressed in Chinese hamster ovary cells and carrying a C-terminal hexahistidine tag, which the authors state they used for every pharmacokinetic and pharmacodynamic study in the paper, and there the pharmacology paper found that molecule poorly suited to systemic administration (Datta-Mannan, 2013).

It is also not a small peptide: follistatin is a glycoprotein of roughly 38 kilodaltons carrying two N-linked glycosylation sites and the four contiguous domains described in the 1988 sequencing paper, which is a different manufacturing and analysis problem from a short chain. If you are reading a certificate for one, what an HPLC purity number does not cover matters more here than usual, and the identity methods on how to read a peptide COA are the part that speaks to whether the right molecule is present at all. The physical form question, what a lyophilized peptide actually is, is separate again.

Is follistatin banned in sport?

Yes, and the ban runs year round rather than only on competition days. The 2026 World Anti-Doping Code International Standard Prohibited List names follistatin on page 11, in section S4.3, agents preventing activin receptor IIB activation, as an example of a myostatin-binding protein alongside myostatin propeptide. The document states that substances in classes S4.3 and S4.4 are non-specified substances, and it came into effect on 1 January 2026.

The gene route is covered by a separate clause: section M3.1 of the same document prohibits the use of nucleic acids or nucleic acid analogues that may alter genome sequences or gene expression by any mechanism, naming gene transfer technologies explicitly. The AAV vectors in the trials above land in that method class rather than in the substance class.

Neighbours on the same list, filed apart: the identical document places ibutamoren, the compound sold as MK-677, at S2.2.4 among growth hormone releasing factors, and GW501516, marketed as cardarine, at S4.4.1 among metabolic modulators. Three sections, three mechanisms. The list prints no reasoning for where it draws those boundaries and we do not supply any.

What the record does not settle

The gaps we can name on this molecule:

• Isoform choice in people. The reasoning for using FS344 over the FS288 route is stated in a review and rests on rat pituitary and cell-culture data. No human trial in the reference list compared the two isoforms head to head.
• Systemic exposure in people. The pharmacology paper did measure circulating FST315 in mice, after single intravenous and subcutaneous doses, putting subcutaneous bioavailability at roughly 90 percent alongside rapid clearance, and it ran the protein through a mouse muscle injury model where daily subcutaneous dosing for seven days did not significantly increase gastrocnemius mass or body weight while daily local intramuscular dosing raised muscle mass by about 14 percent. Then it engineered around the problem. What no source here settles is what circulating follistatin protein does in a person, because every human record on this page is local gene transfer.
• Size of the human record. Six Becker patients and six myositis patients treated in the published reports, sitting inside one registry record whose actual enrollment is fifteen and which posts no participant flow of its own, plus three Duchenne participants in a second record whose posted flow shows three started and three completed, all by intramuscular AAV gene transfer, and the two published studies used matched or internal comparators rather than randomization. A third completed record, whose intervention is a follistatin plasmid instead of a virus, carries an actual enrollment of 43 and posts neither results nor participant flow, and two further follistatin gene therapy records, NCT07443826 and NCT07285629, were still recruiting when we checked. More people have been enrolled than have been reported on, and the gap is not small.
• Marketed product. A DailyMed search for the drug name follistatin, run on 2 August 2026 against a database published 31 July 2026, returned zero records, and an openFDA drug label query for the same substance name returned no matches. Both of those are label repositories, which hold marketed labeling rather than approval decisions, so what this establishes is that neither repository carried a label under that name on that day, and nothing about what any regulator has or has not approved.
• The affinity conflict. A tenfold difference in activin affinity between FS315 and FS288 is asserted in one review and not reported by either primary paper we retrieved. That is unresolved on this page.

How this page is sourced

Primary documents, read directly: the sequence and isoform biology come from the papers below, each abstract or record retrieved from PubMed on 2 August 2026; the residue and glycosylation annotations from the UniProt entry itself; and the three trial records, plus the registry-wide intervention search recorded in reference 18, from the ClinicalTrials.gov v2 API.

Two documents we could not get from their publishers: the anti-doping classification is read at page 11 of the official 2026 Prohibited List, but on 2 August 2026 every request to the World Anti-Doping Agency's own link returned an empty bot challenge instead of the file. The copy read for this page is the one hosted by the International Testing Agency under the same filename, and reference 13 records its size and checksum. The 2013 pharmacology paper behaved the same way: on 3 August 2026 the publisher's article page and its PDF both returned a bot challenge, so the version of record was read from the Internet Archive's capture of that same publisher PDF, whose size and checksum reference 10 records. In neither case can we compare the copy we read against the publisher's own live file, because the publisher's own live file is what would not download.

What is a review and what is not: two entries below are review articles rather than primary measurement, and both are named as reviews where they are cited. Where a review and a primary paper disagree, this page prints both rather than choosing, which is why the affinity conflict is still open in the section above.

Eighteen sources: eleven papers with PubMed IDs, one protein sequence database entry, one international standard, three clinical trial registry records, one registry-wide search of that same registry, and one drug label repository query.

The standard: how we pick sources, how we verify an identifier, and who reads a page before it ships are all set out on the Decadewise methodology page.

Last reviewed: 2 August 2026.

1. 1

   Shimasaki S, Koga M, Esch F, et al. Primary structure of the human follistatin precursor and its genomic organization. Proc Natl Acad Sci U S A. 1988;85(12):4218-4222. doi:10.1073/pnas.85.12.4218. PMID 3380788.

2. 2

   Sugino K, Kurosawa N, Nakamura T, et al. Molecular heterogeneity of follistatin, an activin-binding protein. Higher affinity of the carboxyl-terminal truncated forms for heparan sulfate proteoglycans on the ovarian granulosa cell. J Biol Chem. 1993;268(21):15579-15587. The PubMed record carries no DOI for this article. PMID 8340384.

3. 3

   Saito S, Sidis Y, Mukherjee A, Xia Y, Schneyer A. Differential biosynthesis and intracellular transport of follistatin isoforms and follistatin-like-3. Endocrinology. 2005;146(12):5052-5062. doi:10.1210/en.2005-0833. PMID 16150905.

4. 4

   Sidis Y, Mukherjee A, Keutmann H, Delbaere A, Sadatsuki M, Schneyer A. Biological activity of follistatin isoforms and follistatin-like-3 is dependent on differential cell surface binding and specificity for activin, myostatin, and bone morphogenetic proteins. Endocrinology. 2006;147(7):3586-3597. doi:10.1210/en.2006-0089. PMID 16627583.

5. 5

   Lee SJ, McPherron AC. Regulation of myostatin activity and muscle growth. Proc Natl Acad Sci U S A. 2001;98(16):9306-9311. doi:10.1073/pnas.151270098. PMID 11459935.

6. 6

   Kota J, Handy CR, Haidet AM, et al. Follistatin gene delivery enhances muscle growth and strength in nonhuman primates. Sci Transl Med. 2009;1(6):6ra15. doi:10.1126/scitranslmed.3000112. PMID 20368179.

7. 7

   Mendell JR, Sahenk Z, Malik V, et al. A phase 1/2a follistatin gene therapy trial for becker muscular dystrophy. Mol Ther. 2015;23(1):192-201. doi:10.1038/mt.2014.200. PMID 25322757.

8. 8

   Al-Zaidy SA, Sahenk Z, Rodino-Klapac LR, Kaspar B, Mendell JR. Follistatin Gene Therapy Improves Ambulation in Becker Muscular Dystrophy. J Neuromuscul Dis. 2015;2(3):185-192. This entry is a review article. It is the inline source for the short answer's wording that FS344 undergoes post-translational modification to FS315, a relationship reference 7 states in its own introduction as well, and it is the only source on this page for the tenfold affinity claim discussed above. doi:10.3233/JND-150083. PMID 27858738.

9. 9

   Mendell JR, Sahenk Z, Al-Zaidy S, et al. Follistatin Gene Therapy for Sporadic Inclusion Body Myositis Improves Functional Outcomes. Mol Ther. 2017;25(4):870-879. doi:10.1016/j.ymthe.2017.02.015. PMID 28279643.

10. 10

    Datta-Mannan A, Yaden B, Krishnan V, Jones BE, Croy JE. An engineered human follistatin variant: insights into the pharmacokinetic and pharmocodynamic relationships of a novel molecule with broad therapeutic potential. J Pharmacol Exp Ther. 2013;344(3):616-623. The word pharmocodynamic is misspelled in the published title and is reproduced here as the record carries it. The article is not open access: Unpaywall reports no open location for the DOI, Europe PMC marks its only full-text link subscription required, and on 3 August 2026 the publisher's own article and PDF pages returned an HTTP 403 bot challenge rather than the file. The version of record read for this page is the Internet Archive capture of the publisher's own PDF, dated 20 July 2018: HTTP 200, application/pdf, 210,730 bytes, eight pages, SHA-256 94de215a989b6f4134334fa4d2f210295d33fa7e153ca796e996b6366e2809f8. Table 2 sits on printed page 619, the Materials and Methods construct description on page 617, and the pharmacokinetic and pharmacodynamic Results passages on page 620, and all three were read from that copy. doi:10.1124/jpet.112.201491. PMID 23249626. archived publisher PDF.

11. 11

    Shi L, Resaul J, Owen S, Ye L, Jiang WG. Clinical and Therapeutic Implications of Follistatin in Solid Tumours. Cancer Genomics Proteomics. 2016;13(6):425-435. This entry is a review article and is cited here only for the statement about which isoform predominates. doi:10.21873/cgp.20005. PMID 27807065.

12. 12

    UniProt Consortium. UniProtKB entry P19883, FST_HUMAN, Follistatin. Sequence length 344, computed mass 38,007 Da, signal peptide 1 to 29, chain 30 to 344, N-linked glycosylation at residues 124 and 288, and two annotated isoforms named FS315 and FS288. Retrieved 2 August 2026. uniprot.org, entry P19883.

13. 13

    World Anti-Doping Agency. World Anti-Doping Code International Standard: Prohibited List 2026, section S4.3 on page 11, section M3.1 on page 14, section S2.2.4 on page 8 and section S4.4.1 on page 11. Effective 1 January 2026. On 2 August 2026 the publisher's own link to this file, at wada-ama.org, returned HTTP 202 with an empty body and an Amazon Web Services bot-challenge header rather than the document, so the copy read here is the International Testing Agency mirror of the same filename: HTTP 200, application/pdf, 382,029 bytes, SHA-256 f0b3fcb0dc48477868983eaddac08c132b5c079d941eaa8d40aa6ff9064ece5a. ita.sport, 2026 Prohibited List PDF.

14. 14

    ClinicalTrials.gov. NCT01519349, Follistatin Gene Transfer to Patients With Becker Muscular Dystrophy and Sporadic Inclusion Body Myositis. Phase 1, enrollment 15, status completed, intervention rAAV1.CMV.huFollistatin344. Record retrieved through the ClinicalTrials.gov v2 API on 2 August 2026. clinicaltrials.gov, NCT01519349.

15. 15

    ClinicalTrials.gov. NCT02354781, Clinical Intramuscular Gene Transfer of rAAV1.CMV.huFollistatin344 Trial to Patients With Duchenne Muscular Dystrophy. Phase 1/2, enrollment 3, status completed. The intervention field of this record reads rAAV1.CMV.huFollistin344. Record retrieved through the same API on 2 August 2026. clinicaltrials.gov, NCT02354781.

16. 16

    ClinicalTrials.gov. NCT06411366, Phase I: Safety and Efficacy of an Injectable Follistatin Plasmid Gene Therapy in Humans. Phase 1, open label, single dose, healthy volunteers, actual enrollment 43, status completed, actual start 18 August 2022 and actual completion 31 August 2023. The intervention is recorded as a genetic intervention named Follistatin plasmid and described as Follistatin-344 plasmid, with the registry summary calling a plasmid a circular piece of DNA used as a non-permanent and non-heritable method of transferring genes. The record carries no results. Record retrieved through the ClinicalTrials.gov v2 API on 2 August 2026. clinicaltrials.gov, NCT06411366.

17. 17

    U.S. National Library of Medicine, DailyMed. Drug name query for follistatin through the DailyMed v2 services API on 2 August 2026, against a database published 31 July 2026: zero records returned. The parallel openFDA drug label query on substance name FOLLISTATIN returned no matches the same day. A label repository is a record of marketed labeling, not of approval status. dailymed.nlm.nih.gov, drug name query.

18. 18

    ClinicalTrials.gov. Registry-wide intervention search, query.intr follistatin, run through the ClinicalTrials.gov v2 API on 3 August 2026: HTTP 200, 14 studies returned, with NCT01519349 and NCT02354781 among them as positive controls. Five of the fourteen are human follistatin gene transfer studies. Three are completed, NCT01519349, NCT02354781 and NCT06411366, with actual enrollments of 15, 3 and 43, which is the basis for the statement that 43 is the largest enrollment among completed follistatin gene transfer records. Two were recruiting on that date: NCT07443826, phase 1/2, estimated enrollment 12, interventions recorded as AAV9-Follistatin gene therapy and a VEGF plasmid; and NCT07285629, early phase 1, estimated enrollment 30, intervention recorded as follistatin and klotho gene therapy. The other nine records measure follistatin or a follistatin-like protein as a blood marker, or test a drug, a supplement, exercise, diet or surgery with follistatin as an outcome, and none of them delivers a follistatin gene. This entry is a record of a search, not a study: it establishes what this registry returned for that exact query on that date, and nothing beyond it. clinicaltrials.gov, intervention query.

Related pages

Around this page: five neighboring pages that cover the classification and identity questions next to this one.

• SARMs vs peptides: the molecule-class split, and why the retail shelf and the pharmacology do not line up.
• Is ostarine a steroid: the same kind of naming question, aimed at the androgen receptor side.
• Is MK-677 a SARM: another compound whose retail label and its anti-doping filing point at different mechanisms.
• How to read a COA: what a certificate of analysis tests, which is the identity question a protein this size makes harder.
• HGH vs peptides: the growth hormone axis, the other route people reach for when the goal is muscle.

The briefing: isoform and classification reads like this one ship in The Decadewise briefing before they land anywhere else.