Vitamin C, Collagen & Why Sailors Fell Apart
Scurvy is not really “a vitamin deficiency.” It is a structural collapse. Your body cannot build the rope that holds it together. Collagen — a third of all the protein in you — is spun on an assembly line inside a cell called a fibroblast, and one step on that line has a fatal design flaw: the enzymes that stiffen the rope hold an iron atom that keeps burning itself out. Vitamin C is the reset button. No vitamin C, no reset; no reset, no rope. And because your collagen is constantly being torn down and rebuilt, when the rebuilding stops, the tearing-down does not. Old scars come apart. Gums bleed. Teeth fall out. Press play and watch the line run — then break it.
Try this: let it run on Normal until the fibre is thick and the old scar is sealed — then hit Scurvy (0 mg). Do not watch the assembly line. Watch the scar in the bottom-right corner. It was healed years ago, and it comes apart in front of you.
Live readout
Oral dosing cannot push this gauge past the dotted line, no matter the dose — absorption is saturable and the kidney dumps the rest. Intravenous vitamin C bypasses the gut entirely and reaches millimolar levels; that is a different pharmacology, studied separately.
What's happening
What is real and what is a model. The dose→plasma curve follows published steady-state depletion–repletion data (roughly: 30 mg/day ≈ 9 µmol/L, 100 mg ≈ 56, 200 mg ≈ 70, 1000 mg ≈ 80 — it plateaus). Plasma below 11 µmol/L is the accepted deficiency threshold; 70–80 µmol/L is tissue saturation. Type I collagen really does melt around 41°C when properly hydroxylated and roughly 15°C lower when it is not — i.e. below body temperature. The speeds here are compressed: real collagen turnover takes months, not seconds, and real scurvy takes 1–3 months of near-zero intake. The plasma→hydroxylation curve, the tissue-integrity meters and the "arbitrary units" counter are an illustrative model, not measured clinical values.
The Science in Plain Language
1. Almost every animal makes its own vitamin C. We are one of the broken ones.
A dog does not eat oranges. Neither does a cat, a cow, or a rat — and none of them get scurvy, because they manufacture vitamin C in the liver from ordinary blood sugar. The pathway is four enzymatic steps long, and we still have the first three. What we do not have is the last one: L-gulonolactone oxidase, the enzyme encoded by the GULO gene.
In humans the GULO gene is still there, sitting on chromosome 8 — but it is a pseudogene. It is riddled with deletions and stop codons. It is a rusted-out engine block bolted into the chassis. We share this exact broken gene with the other haplorrhine primates (monkeys and apes), and the same enzyme has been independently lost in guinea pigs, some fruit bats, and certain songbirds. Everyone who lost it was, at the time, eating a diet drenched in fruit. The mutation cost nothing — until somebody got on a ship.
The scale of what we gave up is worth sitting with. Animals that still make their own produce it in the range of tens of milligrams per kilogram of body weight per day, and they ramp production up sharply under stress or infection. Scaled to an adult human, that is grams per day, self-regulating, on demand. We replaced it with a shopping list.
2. The Gly–X–Y rule: why glycine has to be every third residue
Collagen is not folded like other proteins. It is a rope — three left-handed chains wound into one right-handed superhelix, about 1,000 amino acids per chain, roughly 300 nm long and 1.5 nm wide. Type I collagen, the kind in your skin, bone, tendon and scars, is two α1 chains plus one α2 chain.
To wind three chains that tightly, something has to sit at the very centre of the twist — and there is almost no room there. Only glycine fits. Glycine's entire side chain is a single hydrogen atom; it is the smallest amino acid there is. So the sequence obeys an almost tyrannical repeat: Gly–X–Y, Gly–X–Y, Gly–X–Y, a thousand residues without a break. Glycine is about a third of the entire protein. In the X position you usually find proline; in the Y position, hydroxyproline (made from proline) or lysine.
How strict is this rule? Swap a single glycine anywhere in that helix for a bulkier amino acid and the rope cannot close. That is not hypothetical — it is the commonest cause of osteogenesis imperfecta, brittle bone disease. One letter. The centre of the twist is that unforgiving.
3. The iron reset button — the step vitamin C actually performs
Here is the part almost nobody is told, and it is the whole story.
After the ribosome spits out the chain, two enzymes go to work: prolyl-4-hydroxylase and lysyl hydroxylase. They bolt an –OH group onto the prolines and lysines. Both belong to a family called Fe(II)/2-oxoglutarate dioxygenases, which means each one holds a single atom of iron in the ferrous state, Fe²⁺, in its active site, and burns molecular oxygen and 2-oxoglutarate (α-ketoglutarate) to do the chemistry.
And the iron is the flaw. Every so often a catalytic cycle goes wrong — the enzyme burns its oxygen without hydroxylating anything — and the iron is left oxidised to Fe³⁺. An Fe³⁺ enzyme is a dead enzyme. It is not damaged, not degraded; it is simply switched off, and it will stay off forever.
Unless something reduces that iron back to Fe²⁺. That is ascorbate. That is the job. Vitamin C is not a building block of collagen — not one atom of it ends up in the rope. It is the reset button on the machine that builds the rope. It is not even consumed in the normal cycle; it is only needed to rescue the enzyme after a misfire. Which is why the amount you need is so small, and why running out is so catastrophic: the enzymes do not slow down gracefully. They go dark, one by one, and stay dark.
(A small delight, since it shows how tightly wound this machine is: the β subunit of prolyl-4-hydroxylase is not a spare part — it is protein disulfide isomerase, a completely separate enzyme doing a completely separate job, drafted in to hold the complex in the endoplasmic reticulum.)
4. Hydroxyproline zips the rope shut — and 37°C is the exam
Why bother adding an –OH at all? Because that one oxygen atom changes the shape and electronics of the proline ring, locking the backbone into exactly the pucker the triple helix requires, and adding a lattice of extra hydrogen bonds and bound water between the three chains. It is the zip.
Properly hydroxylated type I collagen melts at around 41°C. Strip the hydroxyprolines out and the melting point drops by roughly 15°C — to somewhere in the mid-twenties. Read that again, because it is the entire disease in one number: an under-hydroxylated collagen helix is unstable at human body temperature. It does not make a weak rope. It makes no rope. The chains flop apart at 37°C, the cell's quality control spots the unfolded mess, and it is destroyed in the endoplasmic reticulum before it is ever secreted. Watch the animation: the helices form, shimmer, and get dragged into the red blob. Nothing reaches the outside world at all.
5. The reveal: in scurvy, old wounds come back open
Now the part that makes scurvy so strange, and so frightening to the men who had it.
Your collagen is not a finished building. It is a building site that never closes. Enzymes called matrix metalloproteinases are constantly chewing old collagen out of your tissues, and fibroblasts are constantly laying new collagen in. The structure holding you together is a standing wave — a balance between demolition and construction.
Vitamin C stops the construction. It does nothing to the demolition. The MMPs do not need ascorbate. They keep working.
So in scurvy you do not merely fail to build new collagen. You lose the collagen you already have. And the tissue that goes first is the tissue that turns over fastest — which brings us to the thing that terrified sailors more than anything else on the list:
Scar tissue is collagen. A wound you healed twenty years ago is being actively maintained, every day, by fibroblasts laying down fresh collagen in the scar. Stop the supply, and the scar is quietly digested from within. Old, long-forgotten wounds reopen. Sailors' healed battle wounds split back open at sea. Bones that had knitted years earlier came apart at the old fracture line. This is documented in the classic scurvy literature and it is not a metaphor — it is exactly what the animation shows when you set the slider to zero and watch the scar in the bottom-right corner.
The rest of the syndrome falls out of the same logic, tissue by tissue:
- Capillaries leak. Vessel walls are collagen-supported. They become fragile and bleed. The classic sign is perifollicular haemorrhage — a tiny ring of blood around each hair follicle — along with easy bruising and bleeding into muscles and joints.
- Corkscrew hairs. The hair shaft itself deforms and coils, because the follicle that shapes it has lost its collagen scaffolding. It is nearly pathognomonic — if you see it, think vitamin C.
- Gums swell, redden and bleed. Oral mucosa turns over fast, so it fails early. This is the sign that gave scurvy its reputation.
- Teeth loosen and fall out. Not because the tooth rots — because the periodontal ligament, the collagen hammock that suspends each tooth in its socket, is collagen and is being digested. The tooth is structurally sound. The rope holding it is gone.
- Wounds will not heal at all. New scar tissue is nearly pure new collagen. With no hydroxylation, there is nothing to lay down.
Timing, for scale: a saturated adult holds a body pool of roughly 1,500 mg of vitamin C. Signs of scurvy appear when that pool falls below about 300 mg, which on a near-zero intake takes something like one to three months. That is the arithmetic of a long sea voyage.
The cure is correspondingly absurd. In the Sheffield experiment of the 1940s, volunteers kept on a scorbutic diet developed the classic signs — and about 10 mg a day was enough to reverse them. Ten milligrams. A sixth of an orange. That is what the Royal Navy's citrus ration was really buying, and it is why the animation's slider does something surprising if you nudge it up off zero: the enzymes flicker back to life almost immediately.
But note what the Sheffield researchers also found, and what the model reproduces: 10 mg/day abolishes overt scurvy while leaving wound healing measurably impaired. Preventing the disease and running the machinery properly are two different targets.
6. Copper: the cross-link nobody mentions
A triple helix is not yet a tendon. Once secreted and trimmed, the helices line up into fibrils — you can see the 67 nm banding under an electron microscope — but at that stage they are still just stacked, sliding past one another like uncemented bricks.
What welds them is lysyl oxidase (LOX), and lysyl oxidase is a copper enzyme. It takes the lysines and hydroxylysines in the helix, oxidatively deaminates them into reactive aldehydes (allysine), and those aldehydes spontaneously bond to their neighbours — permanent covalent cross-links between molecules. That, and only that, is what turns collagen into something you can hang your body weight from.
Take the copper away and you get a completely different failure from scurvy: the helix forms perfectly, gets secreted perfectly, assembles into fibrils perfectly — and the fibre never sets. It slips. Connective tissue and blood vessels become weak and fragile. This is exactly what you see in copper-deficient animals, and in the human genetic copper-transport disorder Menkes disease. It is also the mechanism of lathyrism, in which a toxin in the sweet pea (β-aminopropionitrile) blocks lysyl oxidase directly.
The practical version of this: the commonest cause of copper deficiency in ordinary people is long-term high-dose zinc. Zinc induces a protein in the gut lining that traps copper and carries it away in shed cells. If you are taking zinc every day at high doses for months, you are quietly running a copper-depletion experiment on yourself. Set the animation to Copper deficiency and watch a failure mode that no amount of vitamin C will fix.
7. The same trick, in your gut: vitamin C and iron
Notice what ascorbate did to the iron in the enzyme: it reduced Fe³⁺ to Fe²⁺. It performs the identical chemistry in your digestive tract, and it matters enormously.
Non-haem iron — the iron in plants, beans, grains and supplements — mostly arrives as Fe³⁺, which is nearly insoluble at the pH of the small intestine and cannot be absorbed. The transporter that pulls iron into your gut wall, DMT1, only takes Fe²⁺. Vitamin C reduces the iron and holds it in a soluble chelate through the pH change, and the effect is not subtle: a modest amount of vitamin C taken with a meal can multiply non-haem iron absorption several-fold, and it partly rescues iron from the phytates in grains and the tannins in tea and coffee that would otherwise lock it up.
This is the single easiest nutrition intervention for a plant-based eater with low iron: squeeze lemon on the lentils, eat the peppers with the beans, and do not drink tea with the meal. (The mirror image: if you have haemochromatosis or iron overload, this is a reason to be careful with high-dose vitamin C alongside iron-rich meals.) There is a whole animation of that mechanism on this site — see iron absorption and hepcidin.
8. The dose myth — and where IV is genuinely different
Vitamin C absorption is saturable, and this is where most supplement advice quietly falls apart.
Up to about 200 mg in a single dose, you absorb essentially all of it. Above that, the intestinal transporters are swamped and fractional absorption falls off a cliff — by around 1,000–1,250 mg you are absorbing roughly half of what you swallow. Meanwhile the kidney, which normally reclaims filtered ascorbate, gives up and starts dumping the excess. The two mechanisms together clamp plasma at a ceiling of roughly 70–80 µmol/L. You can see it on the gauge: drag the slider from 200 mg to 2,000 mg and watch the needle refuse to move.
So, plainly:
- ~10 mg/day prevents scurvy.
- ~100–200 mg/day saturates plasma and tissue. This is what fruit and vegetables deliver without effort.
- The RDA is 90 mg (men) and 75 mg (women), with an extra 35 mg for smokers — a real, formal adjustment, because smoking raises oxidative turnover and smokers run measurably lower plasma at the same intake. Try the Smoker scenario: same dose, lower plasma, no reserve.
- Grams of oral vitamin C mostly produce expensive urine. The tolerable upper intake level is 2,000 mg/day; above that you get osmotic diarrhoea and cramps, and in susceptible people the extra oxalate load raises the risk of calcium-oxalate kidney stones.
Intravenous vitamin C is a different drug. This is not a hedge — it is straightforward pharmacology. Infusing ascorbate bypasses the gut entirely and evades the whole absorption ceiling, reaching plasma concentrations in the millimolar range: one to two orders of magnitude above anything achievable by mouth. At those concentrations ascorbate stops behaving like a vitamin and starts behaving like a pro-oxidant drug, generating hydrogen peroxide in tissue. That is why it is studied separately, mostly as an adjunct in oncology and critical care. Results so far are mixed and it is not established therapy. The honest summary is: the pharmacology is genuinely distinct, the clinical case is unsettled, and neither of those facts is a reason to swallow more tablets.
Which brings us to Linus Pauling, who deserves better than the sneer he usually gets. Pauling was right about the thing that mattered most: that vitamin C is not a trivial nutrient, that the RDA was set to prevent scurvy rather than to optimise anything, and — a point he made against his critics and which turned out to be pharmacologically correct — that the failed oral trials of vitamin C in cancer could not disprove results obtained with intravenous vitamin C, because the two do not reach the same blood levels. He was wrong that megadoses prevent colds in ordinary people (large reviews show regular supplementation does not stop you catching one, though it modestly shortens the illness, and it does appear to help people under extreme physical stress). Being partly right about a hard problem, decades early, is not a disgrace. It is what the front of science looks like.
9. Collagen supplements, honestly — and how much food you actually need
Collagen powder is protein. When you swallow it, your stomach and small intestine take it apart into amino acids and short peptides, exactly as they do to a chicken breast. There is no courier service that carries it, intact, to your face. Anyone who tells you they are "replenishing your skin's collagen" by mouth is describing a route that does not exist.
That said, the honest version is more interesting than a flat debunk. Some di- and tripeptides from hydrolysed collagen — particularly prolyl-hydroxyproline and hydroxyprolyl-glycine — genuinely do survive digestion and turn up intact in the bloodstream, and there is a real hypothesis that they act as signalling molecules that nudge fibroblasts into building more matrix. Several trials of hydrolysed collagen peptides (typically 2.5–10 g/day) report modest improvements in skin elasticity and joint comfort. The evidence is genuinely mixed, the effect sizes are small, and a great deal of the literature is funded by people who sell collagen. It is not nothing. It is also not what the marketing says it is.
And it misses the point of this page entirely. Raw materials are almost never the bottleneck. Glycine, proline and lysine are abundant in any adequate diet. The step that fails is hydroxylation — and no quantity of collagen powder does anything for an enzyme whose iron is stuck as Fe³⁺. Nor, for that matter, does the copper cross-linking step care. Feeding a stalled assembly line more parts does not restart it.
What actually restarts it is boringly cheap. Real amounts, roughly:
- Red bell pepper, ½ cup raw — about 95 mg. The best common source by a wide margin, and better than any citrus.
- Orange, one medium — about 70 mg. Orange juice, ¾ cup — about 90 mg.
- Kiwifruit, one medium — about 64 mg.
- Broccoli, ½ cup cooked — about 51 mg. Brussels sprouts, ½ cup cooked — about 48 mg.
- Strawberries, ½ cup sliced — about 49 mg.
- Potato, one medium baked — about 17 mg. Unglamorous, but potatoes were historically a major supply, and a population that lost them lost its vitamin C.
One bell pepper and one orange, and you are done for the day, at saturation, with change to spare. Vitamin C is water-soluble and heat-labile, so long boiling in lots of water leaches it out — steam, roast, stir-fry, or eat it raw. That is the entire intervention. It cost the Royal Navy a barrel of lemons and it took them two hundred years to believe it.
Connections
- All Interactive Visualizations
- Vitamin C — the full topic page
- Vitamin C & Collagen — in depth
- Vitamin C Deficiency (Scurvy)
- Vitamin C & Iron Absorption
- Where to actually get vitamin C
- Collagen — the protein itself
- Glycine — every third residue
- Proline — the X position
- Lysine — the Y position & cross-link site
- Copper — lysyl oxidase's cofactor
- Iron — the atom vitamin C keeps resetting
- Bell Peppers — the best common source
- Broccoli
- Oranges
- Strawberries
- Visualization: Iron Absorption & Hepcidin
- Visualization: How Your Body Reads DNA
- Visualization: Bone Remodeling & Calcium