TB-500 / thymosin beta-4

Thymosin beta-4 (Tβ4) is a 43-amino-acid peptide found in essentially every cell, and it is one of the better-characterised molecules in this section — not a compound invented for the supplement market. "TB-500" is something narrower. When a commercial TB-500 product was put through high-resolution mass spectrometry, its active ingredient was Ac-LKKTETQ: the seven residues at positions 17 to 23, acetylated at one end. That fragment and the 43-residue protein are different molecules, and nearly all of the research below used the protein.

What thymosin beta-4 does in a cell

It buffers actin. In resting human neutrophils Tβ4 sits at roughly 149 micromolar in the cytoplasm — about 169 femtograms per cell — and holds most of the free actin monomer pool; when the cell is signalled to build filaments, the complex releases within seconds. That is the whole mechanism at the level of chemistry: it shifts the concentration at which actin filaments assemble. Very similar actin-binding domains in the same structural family do the opposite, nucleating and elongating filaments, so this short module can move actin in either direction depending on what it partners with.

The protein's other activities sit on different stretches of the sequence. The four residues at the N-terminus block inflammation and reduce fibrosis. The central seven — the ones in TB-500 — carry the angiogenesis, cell-migration and wound-healing activity. Downstream, Tβ4 forms a complex with PINCH and integrin-linked kinase and switches on Akt, a survival signal, and it induces matrix metalloproteinases in several cell types. Both of those recur later in this entry, in a less welcome context.

The repair literature

Heart. Thymosin beta-4's therapeutic career began in 2004, when a paper in Nature showed it activating integrin-linked kinase and Akt, improving myocyte survival after coronary ligation in mice and leaving cardiac function better. A second Nature paper found Tβ4 essential for coronary vessel development and able to wake quiescent adult epicardial explants into producing fibroblasts, smooth muscle and endothelium. The follow-up record is uneven. Two independent groups tested the regenerative interpretation and could not reproduce it: with genetic lineage tracing, post-infarction Tβ4 thickened the epicardium and increased capillary density but produced no cardiomyocytes from epicardial cells, and a reporter-mouse reassessment found a single progenitor-marked cell across roughly 2,500 sections from eight treated hearts, against two cells in five controls, with infarct size no different between the arms. The blood-vessel effect replicates; the new-heart-muscle effect has not. A 2023 group did recover a benefit, but by co-overexpressing Tβ4 with prothymosin alpha rather than by giving the peptide.

In humans, recombinant Tβ4 has been through a first-in-human phase I — 54 volunteers on single intravenous doses from 0.05 to 25 micrograms per kilogram and 30 more on ten daily doses, mild-to-moderate adverse events, no dose-limiting toxicity, no accumulation — and a randomised, placebo-controlled trial in 96 patients after a ST-elevation heart attack. That trial's headline is a smaller infarct area at 90 days, and the detail matters: the reduction was in the 43 patients dosed within eight hours of their angioplasty, while the overall comparison across all 96 patients was not significant.

Eyes. This is where the largest human dataset sits, and the arc is instructive. A nine-patient trial in severe dry eye reported 35 percent less ocular discomfort and 59 percent less corneal staining than vehicle. A 72-patient trial missed both of its primary endpoints — each running numerically in placebo's favour — and reported significance only from analyses its own results section calls ad hoc: 27 percent less discomfort during an environmental challenge, better central and superior corneal staining. Its abstract concludes that the study confirms the drug's efficacy. A trial labelled phase III in neurotrophic keratopathy healed six of ten treated patients against one of eight on vehicle, which works out to p = 0.066: short of the line, in a study stopped at eighteen patients of a planned forty-six. Its two sources disagree on why — the paper says slow recruitment in a rare disease, the registration says a business decision — and randomization had left the vehicle arm nearly a decade older, with the larger defect. The 2025 Cochrane review of neurotrophic keratopathy scored that result at a risk ratio of 9.0 with a confidence interval running from 0.57 to 142 — an interval that includes no effect — and rated the evidence low-certainty.

Then it was scaled. ARISE-2 randomised 601 patients with dry eye. At day 29 both co-primary endpoints moved fractionally in the placebo group's favour: ocular discomfort rose 0.07 of a point on treatment and fell 0.04 on placebo, corneal staining rose 0.07 against a fall of 0.01, on scales where lower is better. Every confidence interval spans zero and each pair overlaps; the staining measure is posted for roughly a third of the cohort, and no secondary outcome has any data posted at all. Two further trials of the same size class, 317 and 700 patients, completed without posting results or publishing a report.

Skin. In rodents the dermal work is consistent: Tβ4 increased wound contracture and collagen deposition in diabetic mice and accelerated healing in 26-month-old mice. That study is also the one place the TB-500 fragment was tested head to head against the whole protein, and in the aged animals the seven-residue fragment matched it. The human trials are two phase 2s in chronic ulcers, both powered for safety rather than efficacy, and both reported their healing counts to the trial registry. In venous stasis ulcers, 12 of 55 treated wounds closed by day 84 against 4 of 17 on placebo — 22 percent versus 24. In pressure ulcers, 8 of 54 against 3 of 18 — 15 percent versus 17. The published description of these trials is that Tβ4 accelerated healing "by almost a month in those patients that did heal," which is a claim about the healers, not about how many there were. A third, in epidermolysis bullosa, stopped at 30 of a planned 36 patients for lack of eligible patients and expired study drug, and posted its counts: 8 of 22 treated erosions closed by day 56 against 5 of 8 on placebo — 36 percent versus 63, with eight placebo patients too few to call the difference either way.

Tendon, ligament and muscle. A rat medial collateral ligament transected and packed with one microgram of Tβ4 in fibrin sealant healed with better-organised collagen and better biomechanics at four weeks. In a 2026 independent rat Achilles study, systemic TB-500 at 60 micrograms per kilogram per day significantly improved load-to-failure and total histology scores while BPC-157 separated from control only on individual subscores, the two peptides never differed from each other, and running the two together added nothing over either alone. Tβ4 released slowly from an electrospun scaffold added to that scaffold's push toward tendon-lineage differentiation in human stem cells. Endogenous Tβ4 rises early in regenerating muscle and pulls myoblasts toward the injury. Set against that, six months of Tβ4 twice weekly in exercised dystrophic mice produced significantly more regenerating muscle fibres and no change at all in grip strength, cardiac function or fibrosis — a reminder that better histology and better function are separate results.

The 2026 scoping review of marketed peptides in The American Journal of Sports Medicine searched TB-500 against bone, fracture, muscle, tendon, ligament, meniscus and cartilage, found two-thirds of the literature to be animal work and the human studies to be a handful without robust controls, and concluded that these compounds should not be used in place of or alongside orthopaedic standard of care.

Cartilage and the joint

There is no human trial of thymosin beta-4 or TB-500 for any musculoskeletal indication, and none registered. What exists in cartilage points in a consistent direction, and it is not the direction a reader would hope for.

Exposed to Tβ4 at two to four micrograms per millilitre, primary articular chondrocytes increase expression and activation of pro-MMP-9, a matrix-degrading enzyme. In a separate line of work, subjecting cartilage explants to physiological cyclical compression upregulated thymosin beta-4 among the mechanically responsive genes — in the same experimental system where loading raised MMP expression and activation. And in cartilage taken from 50 patients, the thymosin beta-4 pseudogene transcript was upregulated in the damaged tissue, switched on by mechanical strain, and raised TMSB4 expression by soaking up a microRNA that would otherwise restrain it; blocking that axis was protective, and the authors propose it as a target to inhibit. Three groups, three methods, one direction.

The observational human data agree in sign without settling causation. Endogenous Tβ4 is higher in the serum and synovial fluid of knee osteoarthritis patients than in controls, and highest in the most degraded knees. Synovial fluid averages about 145 nanograms per millilitre in osteoarthritis and about 1,360 in rheumatoid arthritis, and in rheumatoid joints it tracks MMP-9, MMP-13, VEGF, urokinase plasminogen activator and two inflammatory interleukins — in osteoarthritic joints, none of them. All of that is consistent with Tβ4 rising because a joint is damaged, and equally consistent with it contributing; a cross-sectional measurement cannot tell them apart.

The nearest thing to a positive joint result comes from the intervertebral disc, which is fibrocartilage rather than articular cartilage. Exogenous Tβ4 significantly reduced apoptosis in human annulus cells stressed with peroxide or serum starvation — while proteoglycan production, tested at a dose 27 times lower, stayed flat on both assay and histology. Gene-delivered Tβ4 let nucleus pulposus cells from a single donor keep growing to a sixth passage where untreated cells stopped at the third, with less apoptosis and senescence at the third. Cells survived better; matrix did not increase.

The risk literature

This is the section that decides the question for most people who read it carefully, and it deserves precision rather than alarm.

What is established. Thymosin beta-4 is overexpressed across a wide range of carcinomas, and more of it goes with worse disease. In 143 colorectal cancer patients, high tumour Tβ4 correlated with lymphovascular invasion, depth of invasion, nodal and distant metastasis and stage, and independently predicted worse recurrence-free and overall survival on multivariate analysis. In models, forcing expression up is causal. Melanoma cells engineered to overexpress Tβ4 grew tumours averaging 21.7 millimetres against 13.3 for controls, and produced an average of 46.7 metastatic lung nodules against 10.9 — a more than fourfold increase — with 2.3 times more cell migration and 4.4 times as many tumour blood vessels. Overexpression drives invasion in colon carcinoma cells with more MMP-7 and less Fas, and patients' liver metastases carry more Tβ4 message than their own matched primary tumours. It triggers epithelial-mesenchymal transition through integrin-linked kinase — the same node through which the cardiac survival effect was reported — and migration through a Rac1 pathway, and it raises proliferation, invasion and MMP-2 in oral squamous cell carcinoma. An unbiased in-vivo CRISPR screen across 30 candidate genes picked thymosin beta-4 out as a driver of diffuse gastric cancer metastasis. Silencing it arrests tumour growth in mice, which is why the oncology field's therapeutic interest in this molecule is in blocking it. The one place the tumour literature touches cartilage directly is a rat chondrosarcoma model, where functional analysis implicated Tβ4 in metastasis — described by its own authors as preliminary.

What is not established. No study tests whether giving thymosin beta-4 or its fragment to a person causes, initiates or accelerates cancer. Every result above is either an expression correlation inside tumours that already exist, or forced overexpression inside tumour cells. Neither is the exposure created by injecting a peptide, and the literature is not even unanimous: a 1993 study whose title reads as though it agrees actually found the reverse, with thymosin beta-4 expressed highly in non-metastatic colorectal lines and weakly in metastatic ones. The human trials reported no cancer signal, but they ran 28 to 84 days and had no power to see one.

The accurate statement is this. The biology TB-500 is sold for — cell migration, angiogenesis, MMP induction, Akt survival signalling — is the same biology the tumour literature implicates, and it has never been measured in humans who took it. That is a serious theoretical concern rather than a demonstrated harm, and it is a reasonable basis for someone with any personal or family cancer history, or an undiagnosed lesion, to leave this one out of a stack.

Two smaller signals belong here for completeness. Injected Tβ4 damaged the intestinal mucus barrier and suppressed autophagy in mice. And deleting Tβ4 from liver stellate cells reduced fibrosis while putting it back restored the scarring — though other groups report the opposite in the same organ, so the liver literature is genuinely split.

Where the material comes from

Thymosin beta-4 is not an approved medicine anywhere. TB-500 is sold as a research chemical, and the one commercial product characterised in the literature contained Ac-LKKTETQ rather than the protein — so vial contents are an open variable, not a given. WADA prohibits thymosin beta-4 and its derivatives, TB-500 named explicitly, under section S2.3 of the Prohibited List, covering growth factors and growth factor modulators: prohibited at all times, in and out of competition, and non-specified, which carries the heavier sanctions. It anchored one of sport's largest doping cases, the Essendon supplements affair, in which 34 players were sanctioned. Detection assays for the acetylated fragment and its metabolites have existed since 2012 and are routine. The orthopaedic literature has begun to name the peptide: a 2026 review in the American Academy of Orthopaedic Surgeons' open-access journal groups TB-500 with the wound-healing peptides, cites preclinical and veterinary tendon and muscle work for it, and states that the field's peptide evidence is dominated by animal models, small cohorts and case series with very limited randomised trials.

Safety

The controlled human safety record covers the protein, not the fragment, at doses far below community practice. Across the phase I volunteer study, the ocular trials and the two ulcer trials, tolerability was unremarkable: no dose-limiting toxicity or serious adverse events over 28 days of observation at intravenous doses up to 25 micrograms per kilogram, two ocular adverse events on drug against five on placebo in the 72-patient dry-eye trial whose abstract reports none, and an adverse-event profile comparable to placebo in the ulcer trials. None of that covers months of milligram-scale subcutaneous injection of a seven-amino-acid fragment, which is what gray-market use means.

Published animal work used one microgram delivered locally in fibrin, or 60 micrograms per kilogram per day systemically, or 150 micrograms twice weekly. The only published dose-response curve is from rat stroke, and it is not monotonic: 2 and 12 milligrams per kilogram improved recovery out to eight weeks, while 18 milligrams per kilogram did no better than control, with a modelled optimum near 3.75. More is not reliably better with this molecule.

The joint-specific caution is the pro-MMP-9 finding in chondrocytes, and it is worth stating plainly for what it is: a short-exposure culture result, and also the most cartilage-specific measurement anyone has made of this peptide. Competitive athletes face a prohibited-at-all-times substance with an established assay.

Open questions

The result that would move this entry is simple to describe and has never been run: thymosin beta-4, or its fragment, put into an osteoarthritic or cartilage-defect joint in a living animal, with cartilage measured. Everything in the joint column today is either a cell in a dish or a ligament and tendon outside the joint capsule.

Beyond that: whether the two unreported dry-eye trials, 1,017 patients between them, ever publish; whether the phase 2 in acute myocardial infarction and the second keratopathy phase 3 now recruiting confirm the small positive signals; and whether Ac-LKKTETQ behaves like the full protein anywhere other than aged mouse skin, the one tissue where the two have been compared directly.

The actin biology is settled, the tendon and ligament results are among the better rodent data in this section, and there is now a controlled human pharmacokinetic and safety dataset to build a joint trial on. The experiment that would move this entry is available to anyone who wants to run it.