What Is Osteocalcin? Bone Protein, Contested Hormone

Bone biology

Osteocalcin spent thirty years as a well-behaved bone protein, was promoted to a hormone in 2007, and in 2020 three independent laboratories deleted its gene and could not find the hormone. The argument that followed is one of the most instructive replication disputes in modern endocrinology, and it is still not settled. It is also a useful case study in how a claim becomes textbook before it becomes certain.

What is osteocalcin?

Osteocalcin — encoded by BGLAP on chromosome 1q22, and also called bone Gla protein or simply OCN — is a small protein made almost exclusively by osteoblasts, the cells that build bone. Stefano Manolagas, writing in PLOS Genetics in 2020, describes it as “a 46 amino-acid protein that is produced and secreted almost exclusively by osteoblasts.” It is among the most abundant non-collagenous proteins in the skeleton.

It was identified in the mid-1970s by two groups working on the same puzzle: Hauschka, Lian and Gallop reported the direct identification of gamma-carboxyglutamate in mineralised tissue in 1975, and Price and colleagues characterised the gamma-carboxyglutamic-acid-containing bone protein a year later. Both papers appeared in PNAS. The finding mattered because gamma-carboxyglutamate — Gla — was known from clotting factors, not from bone.

One detail that becomes important later: humans have a single osteocalcin gene, mice have two adjacent ones, Bglap and Bglap2, and rats have one. Any mouse knockout must delete both, and the genomic neighbourhood is crowded — the BGLAP locus sits next to PMF1, with documented read-through transcription between them.

Vitamin K, three Gla residues, and a pH trick

Osteocalcin carries three gamma-carboxyglutamate residues. Making them requires the enzyme GGCX and, as its cofactor, vitamin K — which is why warfarin, a vitamin K antagonist, increases the proportion of osteocalcin that leaves the osteoblast uncarboxylated.

Those three carboxylated residues chelate calcium, which is what glues osteocalcin tightly to the hydroxyapatite mineral of bone. And here is the elegant part, in Manolagas’s own description: when osteoclasts resorb bone, “the acidic pH in the resorption compartment causes the carboxyl groups on OCN to be removed, and decarboxylated OCN is released into circulation.” Undercarboxylated osteocalcin in blood is therefore, mechanically, a by-product of bone being broken down.

The use nobody disputes

Clinically, serum osteocalcin has long been used as a marker of bone formation and turnover. It has real limitations: it is cleared by the kidney, so it rises in renal impairment for reasons that have nothing to do with bone; it varies across the day; and circulating osteocalcin is a mixture of intact protein and fragments that different assays detect differently. Total osteocalcin and undercarboxylated osteocalcin are not the same measurement, and much of the human literature that gets cited in the hormone debate measured the former.

The 1996 surprise

If a protein is that abundant in bone and binds mineral that tightly, deleting it should weaken the skeleton. In 1996, Patricia Ducy, Gerard Karsenty and colleagues reported in Nature that it did the opposite. Their abstract is worth quoting: the mice “develop a phenotype marked by higher bone mass and bones of improved functional quality,” and “the absence of osteocalcin leads to an increase in bone formation without impairing bone resorption.” By one later account, cortical thickness reached 150% of wild-type. Osteocalcin, it seemed, was a brake on bone formation, not a builder.

2007: the skeleton is promoted to an endocrine organ

The bigger claim arrived a decade later, and by Karsenty’s own account it came from an accident. Mice engineered for a different purpose — Esp knockouts — turned out to have too much osteocalcin, and as he later wrote, they “were not only hyper-osteocalcinemic they were also hyperinsulinemic, hypertestosteronemic, agitated, and had the ability to run much longer than their wild type littermates.”

From that starting point, a large body of work proposed that undercarboxylated osteocalcin is a genuine hormone acting through the receptor GPRC6A: stimulating insulin secretion and adiponectin, driving testosterone production in Leydig cells, supporting muscle during exercise, influencing the developing brain, and mediating the acute stress response. It is a striking idea — the skeleton not as scaffolding but as an endocrine organ talking back to the pancreas, the testis and the brain — and it moved quickly into reviews, textbooks and even the RefSeq database summary for the human gene, which today still states that osteocalcin “regulates bone remodeling and energy metabolism.”

2020: three knockouts, three answers

Then, on the same day in May 2020, PLOS Genetics published two independent osteocalcin knockouts. Neither found the hormone.

Diegel and colleagues at the Van Andel Institute made a CRISPR double-knockout with no detectable osteocalcin in serum. They found real changes in bone chemistry — altered collagen maturity and carbonate-to-phosphate ratio — but micro-CT and three-point bending found no difference in bone mass or strength, and, in their words, “serum glucose levels and male fertility… did not have significant differences from wild-type littermates.” They add, unusually frankly: “We cannot explain the absence of endocrine effects in mice with this new knockout allele.”

Moriishi, Komori and colleagues in Nagasaki reached the same place by a different route. Their conclusion: osteocalcin “is not involved in the regulation of bone quantity, glucose metabolism, testosterone synthesis, or muscle mass.” What they did find is arguably more interesting — the c-axis of the bone’s apatite crystallites, normally parallel to the collagen fibrils, was severely disrupted, and the bone was weaker as a result. Their summary line is blunt: osteocalcin “is required for bone quality and strength… but it does not function as a hormone.”

In a follow-up comment, Moriishi and Komori published the scale of their negative result: glucose tolerance tests in 58 wild-type and 62 knockout mice, insulin tolerance tests in 27 and 24, on normal and high-fat diets, across ages from 11 weeks to 18 months. All normal.

They also offered the most concrete technical explanation on the table. Their mice were backcrossed to C57BL/6N more than eight times; the original work used a mixed C57BL/6J and 129/Sv background. C57BL/6J — but not 6N — carries a mutation in the Nnt gene and shows impaired glucose metabolism. If homozygous breeding drifted that mutation into the knockout line but not the controls, a glucose phenotype could appear from genetics rather than from osteocalcin. They are careful to note this would not explain the testis and muscle findings.

A rat knockout from 2016 adds a third answer. Osteocalcin-null rats had more trabecular bone and better biomechanics — and, on testing, improved insulin sensitivity, with significantly lower blood glucose 30 minutes after a glucose load. That is a metabolic phenotype, but pointing the opposite way from the one the hormone hypothesis predicts.

The reply: what counts as a hormone?

Karsenty’s response, also in PLOS Genetics, does not dispute the new mice. It disputes the question. His argument is definitional: “a hormone is defined both experimentally and physiologically by what it does when present in abundance… As long as a molecule has not been shown to affect a particular physiological function when injected in a normal animal, it may be many things… but a hormone it may not be.” His central objection to both 2020 papers is procedural: “neither of these papers addresses the question of whether osteocalcin is a hormone. They did not inject wild-type animals with any of the many preparations of osteocalcin shown by many other studies to work.”

He points to a literature he describes as coming from “over 30 groups and counting, across 5 continents,” with specialised experiments run in named collaborators’ laboratories, and concedes the base rate honestly: “for every 20 publications or so showing that osteocalcin is a hormone there was always one failing to do so.”

Notably, the Komori group grants him part of the point. Uncarboxylated osteocalcin, they write, “is a biologically active protein, which exerts a hormone-like effect when administered” — they simply argue that pharmacology at high dose is not the same as physiology, and that they set out to test the latter.

The receptor problem

For osteocalcin to be a hormone it needs a receptor, and the proposed one is GPRC6A. That receptor has its own unresolved literature: some groups reported no bone phenotype in GPRC6A knockout mice and maintained metabolic and skeletal homeostasis in a full-locus knockout, while others have built an entire signalling model on it.

The strangest wrinkle is a human-specific sequence difference, laid out by Pi, Nishimoto and Quarles in a 2021 Endocrinology paper written expressly to explain the divergence.

Their own summary of the state of play is candid: the significance of osteocalcin/GPRC6A signalling “has been questioned,” there is “a lack of metabolic abnormalities in newly created genetically engineered” knockouts, GPRC6A may also act as a tumour suppressor, and “explanations for these divergent findings are elusive.” They note the strongest structural argument against a major human role — there is no known inherited monogenic disorder caused by inactivating GPRC6A mutations — and still conclude that the receptor “should remain a potential therapeutic target.” That is roughly where an honest reader should sit.

So what is osteocalcin, today?

Two things are worth separating. The matrix story got better, not worse: both 2020 groups found genuine mineral abnormalities in osteocalcin-null bone, and Moriishi’s crystallite-alignment result is a more precise account of what the protein does than “it binds hydroxyapatite.” The hormone story is unresolved, with the original laboratory and a growing set of independent laboratories reporting incompatible results, and no agreed explanation for the discrepancy.

There is a broader lesson here about how much weight a single mouse line can carry, and it connects to something we have written about before in why peptide trials are small: a striking finding in an animal model is a hypothesis, not a fact about humans. Osteocalcin got as far as reference-database summaries and popular supplement marketing on the strength of mouse data that has not consistently replicated. That should make anyone cautious about the next skeleton-talks-to-the-brain headline.

Frequently asked questions

Is osteocalcin a hormone or not?

It depends on which definition you accept and which experiments you weight. Injected at pharmacological doses, undercarboxylated osteocalcin produces effects that even its critics acknowledge. Deleted from the genome, three independent 2020-era models showed normal glucose metabolism, testosterone and muscle. Those two statements can both be true, and that is precisely what the argument is about.

Does taking vitamin K change my osteocalcin?

Vitamin K is the cofactor for the enzyme that carboxylates osteocalcin, so vitamin K status changes the ratio of carboxylated to undercarboxylated protein — that part is straightforward biochemistry. Whether shifting that ratio produces meaningful metabolic effects in humans is a separate question, and the trial literature is mixed. This is a question for a clinician, not for a supplement label.

Why does my lab report list osteocalcin as a bone marker?

Because that use predates the hormone debate by decades and is not affected by it. Serum osteocalcin reflects bone formation and turnover. Interpretation still requires care around kidney function, time of day, and which assay was used.

What is the single strongest piece of evidence against the hormone hypothesis?

Arguably not the metabolic data at all, but the skeletal data. The 1996 knockout reported markedly increased bone mass; both 2020 knockouts reported normal bone mass. The groups disagree about the most basic, most easily measured phenotype in the animal — and that disagreement has to be resolved before the downstream endocrine claims can be adjudicated.

Informational only — not medical advice · 21+. Nothing here is dosing, supplement or diagnostic guidance. Decisions about vitamin K, bone health or laboratory testing belong with a qualified healthcare professional.