BMP-7 is usually described as "an anti-fibrotic factor". That is only half true, and it misleads precisely because it is half true. This molecule has a feature almost none of its neighbours in the tissue-repair literature share: it really was a product you could buy. Not from a research catalogue, from an operating room supply list. And then it was taken away.

The molecule you could once actually order into an operating room

First, the class, because this is where people go wrong. BMP-7 (bone morphogenetic protein 7) is a growth factor of the TGF-β superfamily, a secreted signalling protein. Not a peptide. The active species is a disulfide-linked, glycosylated homodimer of roughly 35 kDa; a single chain on its own is inactive. This is not a terminological quibble: everything follows from it, from how the material is manufactured to what belongs on a certificate of analysis.

Under its second name, OP-1, osteogenic protein-1, the same protein existed for more than a decade as a medical device. Recombinant human BMP-7 loaded onto a collagen carrier was authorised by the FDA (U.S. Food and Drug Administration) under a humanitarian device exemption: in 2001 for long-bone nonunions, later for revision spinal fusion. The registration study in tibial nonunions compared it against autograft, the gold standard of bone grafting.

And this is where it gets interesting. BMP-7's clinical career happened in orthopaedics. Its scientific fame was earned in nephrology. The two lines barely intersected.

Why "brake" is more accurate than "growth factor"

The standard description says BMP-7 counteracts fibrosis. Mechanistically it is more accurate to put it differently.

TGF-β1, the central profibrotic signal, works through Smad2/3. BMP-7 works on the same receptor principles but through a different arm: Smad1/5/8. Both arms compete for the shared partner Smad4 and mutually repress one another. These are not two independent processes but one lever with two ends.

Now the key point. In the healthy kidney, BMP-7 is produced constitutively. After injury, ischaemic, toxic, metabolic, its expression falls. Fibrosis proceeds not because a new stimulus switched on, but because restraint switched off.

That fundamentally changes what "giving BMP-7" means. You are not adding a repair signal, you are trying to restore a brake that disappeared. And "will this prevent new scar" and "will this dissolve existing scar" are two different questions with different answers. A large share of the confusion around this molecule grows from exactly here.

What the kidney models showed, and where they stopped

The preclinical data here are genuinely solid and should not be minimised. In rats, OP-1 reduced injury severity in ischaemic acute renal failure. In a ureteral obstruction model, it prevented fibrogenesis. The loudest paper appeared in Nature Medicine in 2003: BMP-7 counteracted TGF-β1-induced epithelial-to-mesenchymal transition in renal tubular cells and, in mouse models, reversed established chronic injury.

Now the limits, without which that picture is incomplete. First, these are rodents, and the doses were supraphysiological. Second, the role of epithelial-to-mesenchymal transition in adult kidney fibrosis itself became the subject of a long professional debate after 2003, it is assessed far more conservatively today. Third, there is a layer mentioned less often: tissue contains endogenous BMP antagonists, USAG-1, gremlin, noggin. They bind BMPs extracellularly, and injury often increases them. So "adding ligand" and "releasing the brake" are not the same thing: part of the delivered protein is neutralised before it ever reaches a receptor.

How an authorised product left the market

This story is easy to misstate in either direction, so precision is worth the effort.

OP-1 never held a full PMA approval, it lived under an HDE, a pathway for devices intended for small patient populations, with lower efficacy-evidence requirements. In 2009 an FDA advisory committee voted against expanding the indication to uninstrumented lumbar fusion. In 2010 Stryker Biotech sold the OP-1 assets to Olympus. In 2014 Olympus wound the effort down and closed its US operations, after which rhBMP-7 was no longer marketed in the United States.

What does not follow: that BMP-7 was found unsafe. The market exit was a commercial and regulatory outcome, a narrow indication, a restrictive authorisation route, a refused expansion, a change of owner. Not a conclusion about harm.

One more distinction worth holding onto: the prominent safety controversy over BMPs in spinal surgery concerned primarily rhBMP-2, a different protein, a different manufacturer, a different product. The two stories are constantly conflated, and they should not be.

The attempt to bypass the protein, and how it ended

The main technical obstacle was never the biology; it was delivery. BMPs bind extracellular matrix avidly, are poorly retained where you put them, and their receptors are near-ubiquitous, so a systemic dose sufficient for the kidney also hits bone. That is exactly why OP-1's clinical form was a solid collagen implant rather than an injection.

The logical workaround: don't deliver the protein, build a small molecule that activates the same receptor. In 2012 it was shown that the ALK3 receptor is critical for kidney regeneration and that a peptide ALK3 agonist (THR-123) reversed established fibrosis in mice. From that came the clinical candidate THR-184.

The result deserves to be stated plainly. In a randomised, placebo-controlled phase 2 trial in patients at high risk of acute kidney injury after cardiac surgery, THR-184 across a range of doses did not reduce the incidence, severity or duration of AKI compared with placebo. AKI occurred in roughly 74–79% of patients in all groups, placebo included. Safety was comparable.

So as things stand: an elegant mechanism, reproducible rodent data, and one properly conducted human trial of the approach, negative. No approved BMP-7-based anti-fibrotic product exists. That is not a reason to write the field off, but it is also not a licence to describe it as proven.

What this means for laboratory work

If you work with BMP-7 as a reagent, a few practical consequences:

  • The dimer decides everything. Activity resides in the disulfide-linked homodimer. The reduced monomer is inactive, the most common reason a reagent "does nothing".
  • The expression system is not a detail. Mammalian-cell material is glycosylated; E. coli material is not and requires refolding. They are not interchangeable in bioassays by default.
  • Mature domain or precursor. BMP-7 is synthesised as a proprotein; the specification should say which unambiguously.
  • Adsorption to plastic. At low concentrations the actually available concentration can differ substantially from the nominal one.
  • Search under both names. Half the relevant literature is filed under "OP-1", not "BMP-7".

A detailed treatment of the mechanism, the evidence base and the analytical requirements is in the BMP-7 monograph. Related material: HGF, unlike BMP-7 a reparative signal rather than a brake, with a very different failure story; Ac-SDKP, a peptide inhibitor of the same Smad2 arm. Catalogue section: tissue repair.

BMP-7 is supplied strictly as a research-use-only (RUO) reagent. It is not a medicine and not a product for human consumption. The material is not intended for diagnosis or treatment.

Related: BMP-7 (OP-1): what it is, the mechanism and an honest review of the research.