TMG (Betaine)

Betaine is trimethylglycine: glycine carrying three methyl groups, one of which it can give away. The enzyme betaine-homocysteine methyltransferase uses it to remethylate homocysteine back into methionine, and methionine becomes S-adenosylmethionine, the molecule almost every methylation reaction in the body draws on. Betaine is also an osmolyte — cells accumulate it to balance osmotic pressure without disturbing how their proteins fold, which is why the kidney concentrates it and why blood and urine betaine behave in ways a simple nutrient would not.

What betaine is taken for: the methionine cycle

Most people who take TMG take it for methylation, usually alongside an NAD+ precursor, and this is the part of the literature that is genuinely well measured — in a biomarker rather than an outcome, but measured carefully and repeatedly.

The dose-response is the useful finding, because it covers the range people actually swallow. In four groups of nineteen healthy adults given 1.5, 3 or 6 g/day or placebo for six weeks, fasting homocysteine finished 12%, 15% and 20% below placebo. The rise after a methionine load — the post-meal spike, which is the harder test — was blunted by 23%, 30% and 40%. A second escalating-dose study is more conservative on size and agrees on shape: nothing detectable at 1 g/day, 10% at 3 g, 14% at 6 g. Pooled across five randomised trials the figure is a reduction of 1.23 µmol/L. All of that is in people whose homocysteine was normal. In a hundred Chinese adults who actually had it elevated, a daily tablet carrying 1 g of betaine alongside folate, B6, B12 and zinc lowered homocysteine 3.87 µmol/L against placebo over twelve weeks — with the fall tracking rises in both blood folate and blood betaine, so the betaine's own share of it is not separable.

Two caveats sit inside that curve. Six months of 4 g/day in young, lean, healthy volunteers whose homocysteine was normal to begin with changed nothing at all, which is what a repletion effect looks like when there is nothing to replete. And betaine does not substitute for folate: in five hundred adults, folate was the stronger determinant of fasting homocysteine while betaine dominated the rise after a methionine load, and betaine mattered most in people whose folate status was low. The two cooperate. Adding a milligram of folic acid on top of 6 g of betaine buys a further five percentage points.

For context on where a capsule sits: cereals supply about two-thirds of dietary betaine in Western diets, whole wheat is the best common source and quinoa the richest measured, and average intake is roughly 131 mg a day. A gram-and-a-half capsule is a step change, not a top-up.

Liver, body composition and performance

The largest properly controlled clinical trial of betaine was negative. Fifty- five patients with biopsy-proven fatty liver disease took 20 g/day for twelve months; the activity score and fibrosis stage did not improve, and betaine failed to lower S-adenosylhomocysteine or the cytokines the mechanism predicted it would. A 2026 dose-finding follow-up found the more modest version of the result: liver enzymes and serum fibrosis markers fell at 8, 4 and 2 g/day, but not at 1 g.

The body-composition literature is small and consistent in direction: pooled across six trials, fat mass fell 2.53 kg and body fat percentage 2.44%, while weight and BMI did not move. Performance is more mixed. A 2024 meta-analysis of seventeen chronic-supplementation trials found improved maximal strength, carried entirely by the lower body, with confidence intervals that nearly touch zero; six weeks in trained CrossFit athletes was null between groups, and head to head against creatine, betaine raised neither muscle phosphorylcreatine nor strength and added nothing when stacked on top of it. Inflammatory markers are essentially flat across six trials.

The homocysteine-to-cartilage argument

Two threads connect the methionine cycle to the joint, and both are about homocysteine being harmful rather than betaine being helpful.

The first is matrix chemistry. Homocysteine suppresses lysyl oxidase, the enzyme that cross-links collagen, working through IL-6 and DNA methylation in bone-forming cells. In chondrocytes specifically, homocysteine represses the SIRT1/AMPK axis — the same pathway the NMN entry is built on, approached from the opposite direction — producing mitochondrial dysfunction, oxidative stress and apoptosis while raising the matrix-degrading enzyme MMP-13. The same pattern appears in cartilage of animals fed into high homocysteine.

The second is genetics. Mendelian randomization associates genetically higher homocysteine with modestly higher osteoarthritis risk, an odds ratio of about 1.10, with folate appearing protective. The effect is small, pleiotropy is hard to exclude, and the protective exposures it measured were folate and B12 rather than betaine.

The closest anyone has come to testing the pathway in a human joint is not betaine at all but SAMe, the metabolite betaine feeds. Cochrane found four placebo-controlled trials in knee or hip osteoarthritis and could pool a pain outcome from two of them, in 533 patients: a difference of about 0.4 cm on a 10 cm scale with a confidence interval crossing zero, no benefit on function, trials judged small and of questionable quality, and a recommendation against routine use.

The rodent cartilage studies, and their doses

Three independent groups have now put betaine into a joint model, and the theme across them is consistent: redox and remodelling modulation.

In mice with surgically induced knee osteoarthritis, betaine reduced proteoglycan loss and histology scores, cut MMP-13 and collagen-X-positive cells, and normalised subchondral bone remodelling and aberrant angiogenesis, apparently by blocking osteoclast formation through the ROS and MAPK pathways. In a rat jaw-joint model, betaine improved chondrogenic differentiation of stem cells and those treated cells improved cartilage repair. In a rat cartilage-defect model, betaine bound and downregulated a fibrosis marker raised in damaged human osteoarthritic cartilage, and the repair tissue came back more hyaline and less fibrotic.

The doses are the thing to hold on to. The mouse study delivered betaine as 2% of the drinking water, continuously — an exposure on the order of grams per kilogram of body weight per day, which scales to tens of grams in a person. The rat defect study injected betaine into the joint once a week. The jaw-joint study treated the cells, not the animal. None of the three tested swallowing betaine at a human dose, which is the experiment the entry is waiting on.

Dose & practical notes

Homocysteine trials put 1.5 g/day at roughly a 12% reduction and 3 g at 15% over six weeks, with little reliable effect below about 1 g. The sports-performance literature uses 2.5 g/day for at least a week. The liver trials moved markers at 2 g and above. Folate is the stronger lever on fasting homocysteine and betaine on the post-meal excursion, so the two are complementary rather than interchangeable. No joint dose exists.

Contradicting evidence & limitations

The lipid signal is the finding that constrains high doses, and it is precise. Six g/day for six weeks raised LDL cholesterol by 0.36 mmol/L and triacylglycerol by 0.14, with the total-to-HDL ratio up 0.23 and the effect already visible at two weeks; lower doses moved LDL the same way without reaching significance. Pooling every trial at 4 g/day or more, total cholesterol rose 0.34 mmol/L. Against that, six months at 4 g/day in healthy young volunteers found no lipid change, and the 2026 liver trials saw no LDL movement anywhere from 1 to 8 g/day. The effect is established at 6 g, present in pooled data at 4 g and above, and not demonstrated in the 1.5–3 g range — which is also the range least studied for it.

Homocysteine itself remains a surrogate. Large homocysteine-lowering trials in cardiovascular disease have a long record of moving the biomarker without moving events, and nobody has run the folate-versus-B12-versus-betaine factorial that would show whether lowering homocysteine changes osteoarthritis incidence.

Plasma betaine is also not a simple more-is-better marker. In 475 patients followed after an acute coronary admission, low plasma betaine predicted events in those without diabetes while high plasma betaine predicted them in those with it — a reversal the authors attribute to betaine leaking out of tissues into urine. In nearly four thousand angiography patients, higher betaine predicted major cardiac events only when the gut-microbial metabolite TMAO was also elevated, and lost significance once TMAO entered the model.

Safety

Well tolerated at 3–6 g/day across the homocysteine trials. The dose-dependent rise in total and LDL cholesterol above 4 g/day is the main documented flag. About a third of participants in the 2026 liver trials reported mild, transient gastrointestinal symptoms. Some users notice a fishy body odour from trimethylamine, which is the same pathway that feeds TMAO. That route has been shown directly: labelled betaine fed by mouth turns up as TMAO in mice, and not at all once their gut flora are suppressed — though betaine is a hundredfold poorer precursor for it than choline is. The longest human safety data come from a rare-disease population: a prospective registry followed 130 people with homocystinuria and related remethylation disorders on prescription betaine for a median of 6.8 years, with roughly half of those over ten years old exceeding the recommended 6 g daily maximum, and identified no new risk. No known joint-relevant drug interactions.

What would move this entry

The biochemistry is real and the human dose-response is one of the cleaner curves on this site — it is simply a curve in a biomarker. Three independent rodent groups now report cartilage protection, which is more preclinical convergence than most supplements in this section have. What is missing is a single experiment: oral betaine, at a dose a person could take, with a cartilage or symptom endpoint. Until someone runs it, the methylation case stands on its own evidence and the joint case stands on rodents.