Place · Level 3
Vitamin K2
和 K1 是姐妹 · 把钙送进骨头 · 把钙挡在血管外 · 和 D 是一队的
Story path
Chapter 1
K1 vs K2
K1 vs K2
Vitamin K is a family that shares the same core (2-methyl-1,4-naphthoquinone head) but differs in tail length:
K1 (phylloquinone): single isoprenoid side chain, comes from leafy greens — it's literally part of the chlorophyll synthesis machinery. The liver takes it up to synthesize clotting factors II / VII / IX / X (γ-carboxylated so they can bind Ca²⁺ and start coagulation). Healthy people rarely lack K1 — leafy greens are everywhere.K2 (menaquinones, MK-n): multi-isoprenoid side chain where n is the unit count, commonly MK-4 through MK-13. Sources are fermented foods and gut microbial conversion. K2 routes to bone and blood vessels — the lead character of extra-hepatic vitamin K-dependent proteins (VKDPs).K3 (menadione): synthetic form without a side chain; hepatotoxic, so not usable as a human supplement — only animal feed and research.
So today's discussion of 'K2 prevents osteoporosis and arterial calcification' is almost never about coagulation — that's K1's territory.
K1 (phylloquinone): single isoprenoid side chain, comes from leafy greens — it's literally part of the chlorophyll synthesis machinery. The liver takes it up to synthesize clotting factors II / VII / IX / X (γ-carboxylated so they can bind Ca²⁺ and start coagulation). Healthy people rarely lack K1 — leafy greens are everywhere.K2 (menaquinones, MK-n): multi-isoprenoid side chain where n is the unit count, commonly MK-4 through MK-13. Sources are fermented foods and gut microbial conversion. K2 routes to bone and blood vessels — the lead character of extra-hepatic vitamin K-dependent proteins (VKDPs).K3 (menadione): synthetic form without a side chain; hepatotoxic, so not usable as a human supplement — only animal feed and research.
So today's discussion of 'K2 prevents osteoporosis and arterial calcification' is almost never about coagulation — that's K1's territory.
Newborn K1 injection
Routine intramuscular K1 within 24 hours of birth is one of obstetrics' most universal 'single-shot vitamin' interventions, and a small public health victory.Why it's needed: newborns have minimal hepatic vitamin K storage (the placenta is poorly permeable to K); the gut microbiome isn't yet established and can't produce K2; breast milk is extremely low in vitamin K (~ 1–4 µg/L, far below formula); clotting factors II / VII / IX / X all need γ-carboxylation via vitamin K to work.
Without it — three subtypes of VKDB (vitamin K deficiency bleeding):
Early-onset (within 24 hours of birth): associated with maternal anticoagulant / anti-epileptic use; rare.Classic (1–7 days): GI bleeding, nose bleeds, umbilical stump oozing; overall incidence in unsupplemented infants 0.25–1.7%.Late-onset (1–12 months): the most dangerous — about 50% present as intracranial hemorrhage (ICH), with ~20% mortality and ~40% severe disability among survivors. Almost exclusively in exclusively breastfed infants who didn't receive the shot.
Prevention: the standard practice is a single 1 mg intramuscular K1 injection at birth — CDC, AAP, and WHO all clearly recommend it. It prevents late VKDB essentially 100%. Some countries (Netherlands, Denmark) use a 3-dose oral protocol (birth + 1 week + 1 month) — slightly less effective but needle-free. A trend of parents refusing the injection (worried about preservatives or aluminum in the vitamin) led to multiple US ICH case reports in the 2010s; AAP 2022 re-emphasized the recommendation.
Implication: newborns receive K1 (phylloquinone) because what's needed is hepatic coagulation function, not bone/vascular function. Ordinary adult diets are rich in K1 from greens and almost never deficient — the clearest clinical embodiment of the 'K1 manages coagulation, K2 manages bone and blood vessels' division of labor.
Chapter 2
K2 sources
K2 sources
K2's main sources are narrow — almost all from fermented foods:
Natto (soybeans fermented by Bacillus subtilis natto): MK-7 content ~1000 µg per 100 g — a sharp lead; one spoonful covers about ten times the daily need.Aged hard cheeses (Gouda, Edam, Brie, etc.): a MK-4 + MK-8 + MK-9 mixture, ~50–80 µg per 100 g.Egg yolks: mostly MK-4, ~15–30 µg per egg (higher in pastured hens).Chicken liver, goose liver: MK-4 ~9 µg per 100 g.Grass-fed butter / animal fats: moderate MK-4.
Ordinary leafy greens contain almost no K2 — only K1. Gut bacteria can partly convert K1 into K2 (mainly long-chain forms like MK-10, MK-11), but the conversion is limited and highly dependent on individual microbiome composition, so 'I eat lots of spinach, so I have enough K2' doesn't fully hold.
Regions with high traditional K2 intake include Japan (natto), the Netherlands (hard cheese + butter), France (cheese), and the Nordics (organ meats) — and these regions show epidemiologic correlations between K2 intake and rates of osteoporosis and cardiovascular events.
Natto (soybeans fermented by Bacillus subtilis natto): MK-7 content ~1000 µg per 100 g — a sharp lead; one spoonful covers about ten times the daily need.Aged hard cheeses (Gouda, Edam, Brie, etc.): a MK-4 + MK-8 + MK-9 mixture, ~50–80 µg per 100 g.Egg yolks: mostly MK-4, ~15–30 µg per egg (higher in pastured hens).Chicken liver, goose liver: MK-4 ~9 µg per 100 g.Grass-fed butter / animal fats: moderate MK-4.
Ordinary leafy greens contain almost no K2 — only K1. Gut bacteria can partly convert K1 into K2 (mainly long-chain forms like MK-10, MK-11), but the conversion is limited and highly dependent on individual microbiome composition, so 'I eat lots of spinach, so I have enough K2' doesn't fully hold.
Regions with high traditional K2 intake include Japan (natto), the Netherlands (hard cheese + butter), France (cheese), and the Nordics (organ meats) — and these regions show epidemiologic correlations between K2 intake and rates of osteoporosis and cardiovascular events.
Gut conversion myth
'Gut bacteria convert K1 to K2, so eating lots of greens is enough' — this is partly right, partly wrong, and the most common misconception in the K2 story.Facts: gut bacteria do synthesize K2 — mainly long-chain MK-10, MK-11, MK-12 produced by colonic Bacteroides and E. coli. But absorption is very limited: K2 is fat-soluble and needs bile acids, but most bile acids have already been reabsorbed in the small intestine; the distal gut absorbs K2 poorly, and most bacterial K2 ends up in feces.
How much does this actually contribute: Conly 1995 isotope studies estimated that less than 10% of K status in healthy adults comes from gut bacteria. Long-term antibiotic use further disrupts the microbiome and lowers K2 synthesis, but clinically rarely causes deficiency (because most K still comes from dietary K1). Newborns and infants haven't established gut microbiomes — part of the source of newborn VKDB risk.
So what about greens:
K1 in certain tissues (especially testes, pancreas, vascular wall) can be converted directly to MK-4 by the UBIAD1 enzyme — a meaningful endogenous conversion.But conversion from greens to long-chain MK-7 or MK-9 essentially doesn't happen; that pathway needs bacteria or dietary sources (natto, cheese).Greens alone cannot cover all of K2's functions — especially osteocalcin and MGP, the two systemic Gla proteins that respond more strongly to MK-7.
Practical: to cover both K1 and K2, eat greens (K1) plus several weekly servings of natto, hard cheese, or egg yolks (K2); the traditional combination of greens + egg yolks + butter is more complete than 'salad only'. After antibiotics, you don't need to supplement K2 specifically for a week or two — eating K-containing foods normally to support microbiome recovery is sufficient.
Chapter 3
Bone · osteocalcin
Bone · osteocalcin
Osteocalcin (OC) is the second-most-abundant protein in bone (after collagen), secreted by osteoblasts. It carries 3 glutamic acid (Glu) residues that must be modified into Gla (γ-carboxyglutamic acid) before they can grab calcium.
This step is performed by γ-glutamyl carboxylase (GGCX), with K2 serving as cofactor. The reaction pathway: Glu + CO₂ + O₂ → Gla; K2 is simultaneously oxidized to K2-epoxide, then reduced back by VKORC1 for reuse.
Each Gla residue carries two negative charges; three Gla in a row act like calcium claws, electrostatically pulling Ca²⁺ onto the hydroxyapatite lattice surface. Without K2, GGCX is inactive, and osteocalcin stays undercarboxylated (ucOC) — like a key that hasn't been ground in.
Clinically, the blood ucOC / cOC ratio reflects K2 status. Multiple observational studies show high ucOC correlates with hip fracture risk. MK-7 at 180 µg/day × 3 years reduces vertebral fractures in postmenopausal women — B-tier evidence, consistent in direction but with insufficient sample size.
This is the chemical root of 'supplementing D + calcium but lacking K2 doesn't raise bone density'.
This step is performed by γ-glutamyl carboxylase (GGCX), with K2 serving as cofactor. The reaction pathway: Glu + CO₂ + O₂ → Gla; K2 is simultaneously oxidized to K2-epoxide, then reduced back by VKORC1 for reuse.
Each Gla residue carries two negative charges; three Gla in a row act like calcium claws, electrostatically pulling Ca²⁺ onto the hydroxyapatite lattice surface. Without K2, GGCX is inactive, and osteocalcin stays undercarboxylated (ucOC) — like a key that hasn't been ground in.
Clinically, the blood ucOC / cOC ratio reflects K2 status. Multiple observational studies show high ucOC correlates with hip fracture risk. MK-7 at 180 µg/day × 3 years reduces vertebral fractures in postmenopausal women — B-tier evidence, consistent in direction but with insufficient sample size.
This is the chemical root of 'supplementing D + calcium but lacking K2 doesn't raise bone density'.
Osteocalcin's double role
Osteocalcin was recently discovered to be not just a bone structural protein — it's also a hormone. This is one of the most interesting endocrinology findings of the past fifteen years.Classic role: γ-carboxylated form (cOC) grabs Ca²⁺ to load into the hydroxyapatite lattice — the main K2 storyline.
New role · endocrine hormone (Karsenty lab, 2007+): undercarboxylated osteocalcin (ucOC) released from bone into the bloodstream acts as a hormone on a series of tissues. Target organs include:
Pancreatic β cells: increases insulin secretionAdipocytes: improves insulin sensitivityMuscle: enhances glucose uptake and exercise adaptationBrain: affects memory (strong in animal studies, weak in humans)Testes: increases testosterone synthesis (Karsenty mouse models)
Here there's an irony: 'K2 insufficient → more ucOC → seemingly the hormonally active form', so does K2 supplementation reduce this hormonal effect? The actual answer: under healthy K2 status, osteocalcin carboxylation and release are balanced, and trace ucOC already provides sufficient endocrine signaling; under severe K2 deficiency, the cost of bone structural damage far outweighs the 'extra hormone' benefit.
Clinically observed corresponding phenomena: ucOC briefly rises after resistance training, partly explaining 'exercise improves insulin sensitivity'; in T2D populations, total osteocalcin correlates inversely with insulin resistance (epidemiologic evidence); postmenopausal bisphosphonate use suppresses bone remodeling and lowers osteocalcin, with some studies linking it to mild metabolic disturbance.
These tell us: bone is not a dead rigid scaffold but an active endocrine organ, in dialogue with glucose metabolism, sex hormones, and the brain — K2 is one of the cofactors in that dialogue. This doesn't change the K2 recommendation, but it suggests that focusing on bone density alone may be insufficient; overall metabolism and bone health are coupled.
Chapter 4
Artery · MGP
Artery · MGP
Vascular smooth muscle cells secrete a 'gatekeeper' called matrix Gla protein (MGP), which plays the mirror-image role of osteocalcin: keeping calcium outside vessel walls and preventing arterial calcification.
MGP also requires GGCX + K2 for activation, with γ-carboxylation converting Glu to Gla. Inactive MGP (ucMGP) loses its ability to bind calcium, and calcium deposits in the vessel wall, accelerating atherosclerosis.
Several classic pieces of evidence:
The Rotterdam Study (Geleijnse 2004): 4807 adults 55+ followed for 7–10 years; the top dietary menaquinone-intake quartile vs the lowest had 57% lower coronary mortality and 52% less aortic calcification.The PREVEND and EPIC-NL cohorts observed correlations in the same direction.Warfarin inhibits VKORC1 and prevents K2 from being regenerated; long-term use significantly accelerates vascular calcification — one of its known side effects.
Practical implication: supplementing high-dose D + calcium without K2 risks calcium going to the wrong place — 'leaving bone and ending up in vessels' is a chemically grounded concern. This is the origin of the 'D · K2 · Ca trio' recommendation.
MGP also requires GGCX + K2 for activation, with γ-carboxylation converting Glu to Gla. Inactive MGP (ucMGP) loses its ability to bind calcium, and calcium deposits in the vessel wall, accelerating atherosclerosis.
Several classic pieces of evidence:
The Rotterdam Study (Geleijnse 2004): 4807 adults 55+ followed for 7–10 years; the top dietary menaquinone-intake quartile vs the lowest had 57% lower coronary mortality and 52% less aortic calcification.The PREVEND and EPIC-NL cohorts observed correlations in the same direction.Warfarin inhibits VKORC1 and prevents K2 from being regenerated; long-term use significantly accelerates vascular calcification — one of its known side effects.
Practical implication: supplementing high-dose D + calcium without K2 risks calcium going to the wrong place — 'leaving bone and ending up in vessels' is a chemically grounded concern. This is the origin of the 'D · K2 · Ca trio' recommendation.
Warfarin / DOACs and K
The relationship between warfarin (Coumadin) and vitamin K is one of clinical pharmacology's most classic 'antagonism' cases.Mechanism: clotting factors II/VII/IX/X all need GGCX to convert Glu into Gla residues to function; K is consumed in the reaction into K-epoxide, which must be reduced back to K by VKORC1 (vitamin K epoxide reductase) for reuse. Warfarin precisely inhibits VKORC1, breaking the K cycle, leaving clotting factors uncarboxylated and lengthening clotting time (raising INR).
INR monitoring: a healthy adult INR is ~1.0; warfarin therapy targets are typically 2.0–3.0 (atrial fibrillation, DVT prevention); mechanical valve patients target 2.5–3.5.
The real diet-warfarin problem: it's not 'don't eat greens' — that's a common misconception. The real issue is 'keep intake stable': if K intake fluctuates wildly, INR rides a rollercoaster, raising both bleeding and thrombosis risk. The guidance is to eat roughly the same amount of greens daily (1–2 servings) so the physician can dose warfarin against that baseline; high-K foods (lots of natto, kelp, kale) don't have to be avoided, but sudden large swings should be.
Direct oral anticoagulants (DOACs) changed the game: dabigatran, rivaroxaban, apixaban, edoxaban directly inhibit thrombin (IIa) or factor Xa — they bypass the K cycle, so dietary K doesn't affect drug effect and routine INR monitoring isn't required. This is one of the past fifteen years' anticoagulation revolutions; most warfarin indications have shifted to DOACs.
So: people on warfarin must consult their physician before adding K2 supplements — otherwise INR may drop significantly and thrombosis risk rises; people on DOACs have no K-related food-drug interaction and can use K-rich foods and K2 supplements normally. In other words, first confirm which anticoagulant you're on, then decide how to eat K.
Chapter 5
D + K + Ca team
D + K + Ca team
Putting the three together, you see the same action from three angles:
D (vitamin D): lets the gut absorb more calcium (opens the TRPV6 channel + calbindin)K2: decides where the absorbed calcium goes — activates osteocalcin to send it into bone, activates MGP to keep it out of vesselsCa²⁺: the cargo itself
Missing any one and dosing another heavily becomes risky: D + Ca but lacking K2 means calcium gets absorbed but no one directs it — bone density doesn't rise while arterial calcification accelerates; K2 + Ca but lacking D means the gut never opens the door — absorption is low and serum calcium can only be borrowed from bone; D + K2 but lacking Ca means full engineering crew but no raw material.
Magnesium (Mg²⁺) is the hidden fourth member: both D hydroxylation steps (CYP2R1, CYP27B1) need magnesium as a cofactor.
This is why modern serious guidelines increasingly emphasize overall dietary patterns (greens + fermented foods + fish + nuts and seeds + sunlight) rather than single high-dose supplements. Human biochemistry is a network, not a linear chain — supplementing a non-working link routes water to where it should go, not 'more is better'.
D (vitamin D): lets the gut absorb more calcium (opens the TRPV6 channel + calbindin)K2: decides where the absorbed calcium goes — activates osteocalcin to send it into bone, activates MGP to keep it out of vesselsCa²⁺: the cargo itself
Missing any one and dosing another heavily becomes risky: D + Ca but lacking K2 means calcium gets absorbed but no one directs it — bone density doesn't rise while arterial calcification accelerates; K2 + Ca but lacking D means the gut never opens the door — absorption is low and serum calcium can only be borrowed from bone; D + K2 but lacking Ca means full engineering crew but no raw material.
Magnesium (Mg²⁺) is the hidden fourth member: both D hydroxylation steps (CYP2R1, CYP27B1) need magnesium as a cofactor.
This is why modern serious guidelines increasingly emphasize overall dietary patterns (greens + fermented foods + fish + nuts and seeds + sunlight) rather than single high-dose supplements. Human biochemistry is a network, not a linear chain — supplementing a non-working link routes water to where it should go, not 'more is better'.
Triad dosing in practice
The 'D + K2 + Ca + Mg four-piece' is the hottest combo in the influencer supplement market. How does it translate to actual food + supplement doses?Diet + sunlight first (most people don't need the full supplement set):
D: 10–15 min outdoor sun 2–3 times/week in spring/summer, plus 2–3 weekly servings of fatty fish + egg yolksK2: 2–3 weekly servings of natto / hard cheese / egg yolks plus moderate animal fat (grass-fed butter etc.)K1: daily dark leafy greens 100–200 gCa: dairy + tofu + sardines + greens, 1000–1200 mgMg: pumpkin seeds + almonds + dark greens + whole grains, 320–420 mg
When you need to add supplements (by scenario):
Scenario A — ordinary adult, diet basically adequate, 25-hydroxyvitamin D: The storage form of vitamin D in blood — the number measured to check D status. > 50 nmol/L: usually no supplements needed; in winter, 1000 IU D3/day as insurance is reasonable.Scenario B — 25(OH)D < 50 nmol/L + no fermented foods: D3 2000–4000 IU/day + MK-7 90–180 µg/day + Mg 200–400 mg/day; calcium from food preferred unless diet is genuinely insufficient.Scenario C — postmenopausal woman + borderline bone density + cardiovascular risk: D3 2000–4000 IU + MK-7 180 µg + dietary calcium 1000–1200 mg + Mg 300–400 mg + resistance training 2–3 times/week. This is currently the best-evidenced combination for 'elderly bone health + cardiovascular insurance'.Scenario D — on warfarin: don't self-supplement K2 — consult your physician first, usually with INR adjustment; D + Ca + Mg are still usable.Scenario E — chronic kidney disease (CKD 3-5): don't self-supplement Mg + K + Ca — excretion is impaired, accumulation risk is high; D is usually given in active form (calcitriol) by a nephrologist.
Some traps to avoid:
'¥99 free-shipping D + K2 + Mg + Ca all-in-one tablet': most ingredient doses are inadequate (D often only 400–600 IU, K2 < 50 µg).High-dose calcium 1000 mg without K2 or D: calcium may go to vessels.Only K2 with no exercise and no calcium: no raw material; K2 is just the engineer.
One sentence: the four-piece mechanism is real, but 90%+ of healthy adults on diverse diets with moderate sun don't need to take all of it. Supplement logic is 'fill the gap', not 'stack'.
Chapter 6
MK-4 vs MK-7
MK-4 vs MK-7
K2's chemical name is menaquinone (MK-n), where n is the number of isoprenoid units on the side chain. The supplement market has two main forms:
MK-4 (4 units): mainly in animal tissues, egg yolks, meat, milk; very short half-life of 1–3 hours; clinically requires multiple daily doses (typically 15–45 mg × 3). Japan's osteoporosis treatment guidelines include high-dose MK-4 protocols, but they're expensive and have poor adherence.
MK-7 (7 units): from natto; ~3-day half-life; 90–180 µg once daily maintains stable plasma levels. This is the mainstream supplement form — low cost, stable effect.
So when choosing a K2 supplement:
MK-7 is the first choice — inexpensive, long-acting, good adherence.MK-4 is more appropriate for already-diagnosed osteoporosis patients on a Japanese GLA-OC protocol.K1 supplements: unless there's a coagulation issue, ordinary people don't need to specifically supplement K1.
Warfarin users must first consult their physician before taking any K — any form of K antagonizes warfarin's anticoagulant effect.
MK-4 (4 units): mainly in animal tissues, egg yolks, meat, milk; very short half-life of 1–3 hours; clinically requires multiple daily doses (typically 15–45 mg × 3). Japan's osteoporosis treatment guidelines include high-dose MK-4 protocols, but they're expensive and have poor adherence.
MK-7 (7 units): from natto; ~3-day half-life; 90–180 µg once daily maintains stable plasma levels. This is the mainstream supplement form — low cost, stable effect.
So when choosing a K2 supplement:
MK-7 is the first choice — inexpensive, long-acting, good adherence.MK-4 is more appropriate for already-diagnosed osteoporosis patients on a Japanese GLA-OC protocol.K1 supplements: unless there's a coagulation issue, ordinary people don't need to specifically supplement K1.
Warfarin users must first consult their physician before taking any K — any form of K antagonizes warfarin's anticoagulant effect.
K2 evidence: where we stand
The right stance on K2 supplements is 'mechanism clear, epidemiology consistent, RCTs early but limited' — not a miracle drug, not a gimmick.Strong evidence (mechanism + large cohorts): undercarboxylated osteocalcin (ucOC) correlates with hip fracture risk, consistently across multiple prospective cohorts; undercarboxylated MGP (dp-ucMGP) correlates with cardiovascular events and aortic calcification (Rotterdam, PREVEND, EPIC-NL, etc.); warfarin inhibiting the K cycle and accelerating vascular calcification is a clinically observed clear side effect.
Moderate evidence (small RCTs): Knapen 2013 (3 years, 244 postmenopausal women, 180 µg MK-7/day) showed vertebral height loss reduced by 1.3% and ucOC/cOC ratio improved; Knapen 2015 (same dose × 3 years) showed arterial stiffness index improved; Sato 2002 (Japan, 45 mg MK-4 × 3/day) reduced refracture rates in prior-fracture patients, but this is treatment-dose, not routine supplementation.
Weak or inconsistent evidence: K2 alone for total cardiovascular mortality has strong epidemiologic correlation but large RCTs aren't complete; K2 vs statins / bisphosphonates for osteoporosis treatment is far weaker than drug-grade therapy.
Practical advice:
Diets that don't lack natto / hard cheese / egg yolks don't need K2 supplements.Near-vegan diets + on high-dose D + on calcium supplements: MK-7 90–180 µg/day is reasonable insurance.Postmenopausal women + borderline bone density: K2 + D + Ca under physician guidance.People on warfarin or Coumadin must not self-supplement and must consult their physician first (it antagonizes the drug).
One sentence: K2's role isn't 'solo disease treatment' but 'making sure the calcium you've absorbed goes to the right place' — relatively low-cost insurance, not a miracle.