Collagen types I, II & III

"Collagen" is not one substance. There are twenty-eight types in humans, three of which matter for anything on this site, and the labelling on food and supplements almost never tells you which one you are getting. This entry exists because the mismatch between them is the most common source of confused reasoning about collagen for joints.

The three that matter

Type I is the workhorse — skin, tendon, ligament, bone, and the great majority of collagen in the body. It is what gelatin is made of, what bone stock delivers, what chicken feet are mostly built from, and what nearly every collagen peptide product on the shelf contains.

Type II is the collagen of hyaline cartilage. It is the structural protein of the tissue this entire site is about, and it is a comparatively small share of total body collagen. Its dietary sources are cartilage specifically — chicken sternum, chicken trachea, the joint surfaces in feet — not skin or bone.

Type III accompanies type I in skin, vessels and granulation tissue. It matters here mainly because repair tissue after microfracture is type I and III fibrocartilage rather than the type II hyaline cartilage that was lost, which is why that repair is mechanically inferior.

The type mismatch

Cartilage needs type II. What you can eat is overwhelmingly type I. On a naive building-blocks model, eating type I collagen to rebuild type II cartilage should be roughly as useful as delivering bricks to a glassworks.

Chicken feet make the tension concrete. They are promoted on the logic that toe joints are cartilage and cartilage is type II — but every published characterization of chicken-feet collagen reports type I, resolving into two α-chains confirmed by mass spectrometry. Nobody has ever quantified how much type II a chicken foot contains. By mass it is a type I food with a type II garnish of unmeasured size.

What digestion does to the type

The best mechanistic evidence says ingested collagen does not work as building material at all, which dissolves the mismatch.

When collagen is digested, what reaches the blood intact is not amino acids for reassembly but small peptides — chiefly Pro-Hyp, a proline–hydroxyproline dipeptide. That peptide reaches articular cartilage in animals, is taken up by chondrocytes, and roughly doubles aggrecan expression while tripling glycosaminoglycan deposition.

The control condition in that work is the decisive part: a mixture of free proline, hydroxyproline and a related tripeptide did nothing. The intact dipeptide is the active species.

Pro-Hyp does not care which collagen type it came from. The Gly-Pro-Hyp motif is common to type I and type II alike, and the human absorption study that first identified Pro-Hyp in blood used hydrolysates from porcine skin, chicken feet and cartilage more or less interchangeably.

So if collagen works, it plausibly works as a signal — and a type I source is a perfectly good way to deliver that signal to type II tissue. This is a mechanism-level argument resting on rodent and cell data, not a demonstrated human fact, but it is the most coherent account available.

Where type genuinely is decisive

One case, and it runs the opposite way from the usual claim.

Undenatured type II collagen supplements work at 40 mg — a dose far too small to matter nutritionally. The proposed mechanism is immunological: intact type II epitopes presented to gut-associated lymphoid tissue induce regulatory T cells that dampen immune reactivity against the same epitopes in joint cartilage.

That requires the collagen to be undenatured — the triple helix intact and recognisable. In the foundational animal work, native type II collagen suppressed arthritis and heat-denatured type II collagen did not, despite containing identical amino acids.

The isolated collagen triple helix denatures around 41 °C, dropping to 35 °C once oxidised; collagen within intact tissue is more stable but still gelatinises well below cooking temperatures. A simmering pot destroys it.

This is the practical conclusion of the whole entry: "chicken feet contain type II collagen" is true, and the UC-II evidence still does not transfer to them, because cooking removes the only property that evidence depends on. If you want that mechanism, the supplement is the only route. If you want the Pro-Hyp mechanism, type is largely beside the point and a good stock is a reasonable delivery vehicle.

The types, as a ledger

The same argument in one table. "In the food?" here means a cooked collagen food — a stock, feet, a gelatin — not a capsule. Read it against the evidence column independently: the type that is easiest to eat is not the type with the most direct evidence, and that mismatch is the entry.

The undenatured row is the one worth pausing on. It is absent rather than unmeasured, and the difference is that no future measurement will change it: this is not a gap in what anyone has looked for, it is a thermal fact about the molecule. Everywhere else on this site an absence is provisional. Here it is not.

What nobody has tested

No human trial has compared collagen types head to head for a joint outcome. There is no study establishing that hydrolysed type II outperforms hydrolysed type I, and the largest trial of hydrolysed type II collagen was null on its primary endpoint. The type-specific marketing runs ahead of the evidence in both directions — and a single head-to-head trial, which nobody has yet run, would settle it.