Protein Is Not Just Protein

For years, I’ve liked the idea of eating roughly 70–80% plant-based, with room for animal foods. But why that number? Why those foods? If I am going to spend decades doing something, I would quite like a better explanation than “it sounds healthy.”

Protein is where this gets awkward. We put the number of grams on the front of the packet and behave as if that settles the discussion. Yet those grams contain different mixtures of amino acids, with different effects on nutrition and nutrient signalling. The barcode has a protein total. My cells have considerably more questions.^{1 2}

My position is that amino-acid balance can meaningfully affect health, and changing that balance deserves a place in how we think about ageing. There are human interventions behind that position, including changes in insulin responses, blood lipids and liver fat. The question I want to explore is what pattern makes sense across a lifetime: more of certain amino acids, less of others, and perhaps periods of each.^{3 4 5}

Leo and Longevity’s Protein: the Key to Longevity helped frame the questions for this article. The studies below are the sources for the scientific claims, including human work published after the video.⁶

Start with the humans

#

A 2023 controlled-feeding study put twenty adults through diets with different degrees of methionine restriction or combined methionine and cysteine restriction. The combined restriction lowered measures including LDL cholesterol, insulin and IGF-1, and raised FGF21, a hormone involved in the response to nutrient availability. Methionine restriction alone produced fewer changes. Calories were tailored to participants’ needs, although body weight still changed.³

An earlier sixteen-week randomised trial in adults with obesity and metabolic syndrome found that methionine restriction increased fat oxidation and reduced liver fat beyond the changes seen in controls. Both groups lost weight; the restricted group did not get an additional improvement in the clamp measurement of insulin sensitivity. Useful results, with a reasonably specific address.⁴

Then there are the branched-chain amino acids, or BCAAs: leucine, isoleucine and valine. In a small crossover trial in people with type 2 diabetes, reducing BCAAs while keeping calories and total protein matched lowered meal-related insulin secretion by about 28%, with similar meal glucose. An oral glucose-sensitivity measure improved; the clamp measurements did not. Importantly, the diet still supplied BCAAs above minimum requirements.⁵

These experiments support a more interesting conclusion than “eat less.” The composition of the protein can change the response, even when total protein or prescribed calories are controlled. That is why I take selective restriction seriously.^{3 5}

The human results also vary. An eight-week trial of sulfur-amino-acid restriction in 59 adults found about 1.14 kg more weight loss than the comparison diet, but most conventional risk markers did not separate clearly, and neither did IGF-1 or FGF21. Both groups ate a plant-based diet with some energy restriction.⁷

Here is the boundary I want to set once: these are mostly short trials of metabolism and risk factors. They establish that manipulating amino-acid intake can produce useful human effects. They do not tell us which ratio or schedule will make a person live longest. That is where the rest of this article moves from results into a working hypothesis.

Growth is useful. So is knowing when to stop.

#

Amino acids are building material, but they also carry information about nutrient availability. mTORC1 helps coordinate the response: among other things, it promotes protein synthesis and regulates autophagy, the machinery that breaks down and recycles cellular material. Leucine is one of its nutrient signals.^{2 8}

That makes “build versus repair” a useful way to start thinking. I want the ability to build muscle, heal and replace worn-out components. I also want the maintenance machinery to do its job. These functions overlap: replacing a damaged protein still requires making a new protein. There is no clean border with a little passport booth between growth and repair.²

A fruit-fly experiment illustrates why the mixture matters. Adding essential amino acids to a restricted diet increased reproduction and shortened lifespan. Adding methionine alone restored egg production without the same lifespan penalty. The researchers could partly separate reproduction from longevity by changing the ingredients.⁹

My interpretation is that the attractive target is a useful rhythm of nutrient signals, with enough material for maintenance and adaptation. Permanently trying to suppress growth sounds to me like a questionable plan for owning a body.

Methionine and glycine: an actual connection

#

Methionine is essential. We need it, and we also use it to make S-adenosylmethionine, mercifully abbreviated to SAM, which donates methyl groups in many reactions. Methionine metabolism also connects to cysteine production.¹⁰

Glycine participates directly. An enzyme called glycine N-methyltransferase, or GNMT, transfers a methyl group from SAM to glycine, producing sarcosine. In flies, increasing this metabolic activity restrained an age-related rise in SAM and extended lifespan. Other fly experiments connect this pathway to spermidine production, autophagy and longevity under reduced insulin signalling.^{11 12}

There are direct dietary interactions too: glycine helped rats handle an artificially high methionine intake, reducing the rise in liver methionine and alleviating growth problems. That is a useful detail because glycine’s role is more interesting than simply being “the amino acid that stops building.” Under those conditions, it helped growth recover.¹³

So yes, I think the methionine:glycine balance deserves attention. There is real biochemistry behind it, and experiments show that changing these inputs or their metabolism changes outcomes. I would describe the ratio as one influence on the growth-and-maintenance system, rather than a calibrated repair meter. A dietary ratio, a blood measurement and the rate of an enzyme reaction are different quantities.^{11 13}

The lifespan experiments add weight to the question. A classic study found roughly 30% longer lifespan in male rats on a substantially methionine-restricted diet. In the three-site mouse Interventions Testing Program, glycine supplementation increased pooled median lifespan by approximately 4% in females and 6% in males. Two different interventions, both pointing towards the importance of amino-acid metabolism.^{14 15}

Cysteine deserves a seat at this table too. In mouse experiments, adding it back reversed several responses to methionine restriction, including the FGF21 increase. That fits the human observation that restricting methionine and cysteine together can behave differently from restricting methionine alone.^{16 3}

I find that much more compelling than a story with one villain, one supplement and a checkout button.

What about adding glycine in humans?

#

The GlyNAC studies are relevant: this combination supplies glycine and N-acetylcysteine. A small sixteen-week randomised trial in older adults reported improvements in several biochemical and physical-function measures. A larger, two-week trial in healthy older people found no significant improvement in its primary endpoint, total glutathione. Different populations and durations, and a combination treatment rather than glycine alone.^{17 18}

Increasing substrate availability for maintenance-related pathways is therefore a serious research direction too. I would keep it distinct from deliberately lowering sulfur-amino-acid availability: GlyNAC supplies a cysteine precursor, so these interventions are not interchangeable routes to the same diet.^{17 16}

Lysine is not a tiny repair contractor

#

Lysine matters for normal protein nutrition and connective tissue, including modifications of lysine residues in collagen. But I would not put lysine and glycine on a repair team and methionine and leucine on a growth team. Their jobs are less cooperative with a tidy diagram.^{19 2}

The practical nutritional distinction is useful, though. Humans need nine indispensable amino acids from food. Cereals tend to be relatively low in lysine; legumes supply useful lysine but tend to be relatively lower in sulfur amino acids. Combining plant sources gives us a way to change the mixture while meeting requirements.^{1 20}

There is even a controlled human study of different lysine:arginine ratios. It found little effect on most cardiovascular measures. That makes it a lower-priority lever in my own thinking than the sulfur-amino-acid interventions.²¹

I would like the extra years to contain working muscles

#

This is the part where a longevity argument can become impressively stupid if we forget what we wanted the extra years for.

Leucine can stimulate muscle protein synthesis, including in older adults. Protein and resistance training support muscle adaptation. Those are useful functions to preserve while discussing less growth signalling.^{22 23}

Plant foods can do that job. In a twelve-week training study, habitual vegans and omnivores eating protein-matched diets gained similar amounts of muscle and strength. A controlled crossover study in older adults also found similar integrated muscle protein synthesis on varied, protein-matched vegan and omnivorous diets.^{24 25}

But the word matched is doing work. A 2026 trial in older adults found that adopting a self-selected vegan diet lowered protein intake and appendicular lean mass; resistance exercise mitigated the loss. Removing foods and replacing their nutritional contribution are two separate tasks.²⁶

For orientation, the Nordic adult protein reference intake is 0.83 g/kg/day, with 1.2–1.5 g/kg/day recommended for older adults to support function. A resistance-training meta-analysis estimated a plateau in additional lean-mass gains around 1.6 g/kg/day. These numbers answer different questions; I would use them to check adequacy and training needs before experimenting with restriction.^{27 23}

Could we alternate building and maintenance?

#

This is the hypothesis I find most interesting: can we retain the benefits of repletion while periodically getting the metabolic response to lower availability?

There is a particularly relevant mouse experiment. Researchers alternated four days of methionine-replete food with three days of methionine-free food each week. The animals developed several of the metabolic benefits seen with continuous restriction, including less adiposity and better glucose regulation. In a follow-up experiment, intermittent restriction also preserved more bone structure and strength than continuous restriction.^{28 29}

That is a compelling reason to investigate cycling. It suggests that the restriction period and the recovery period may each contribute something useful. The three methionine-free days were a laboratory mouse intervention, though; I would use the experiment to motivate human research, not copy its menu.

So how would I rank the possibilities?

Every day: meals, then a break

#

My first choice would be ordinary, deliberate meals with enough protein, and an overnight period without eating. I like this as a practical starting hypothesis because it leaves room for adequate nutrition without requiring every waking hour to contain another protein product.

There is human evidence that timing matters. In a small controlled trial in men with prediabetes, an early eating window improved insulin sensitivity and blood pressure without weight loss. That intervention changed the timing of the whole diet, including its relationship to the body clock; it did not isolate amino acids.³⁰

Protein distribution can also affect the muscle response. One exercise study found that four moderate servings over twelve hours supported more muscle protein synthesis than the same total divided into very small frequent servings or two larger servings. Other work shows that a large protein meal can sustain an anabolic response for more than twelve hours. So there is no sensible universal claim that everything above thirty grams is wasted, or that digestion switches off on a tidy timetable.^{31 32}

The idea I would explore is adequate feeding followed by a regular break, rather than chasing the smallest possible eating window. I would judge it partly by whether it still supports training, appetite, sleep and recovery.

I also would not put “repair begins at hour sixteen” on a diagram. A 2026 human trial actually measured autophagic flux in circulating immune cells and found no increase after four weeks of moderate protein restriction. Tissue, energy intake and the specific intervention matter. My daily-rhythm hypothesis is about a plausible pattern to test, not a stopwatch for cellular housekeeping.³³

Different days: vary the mixture

#

Of the more experimental options, lower-sulfur-amino-acid days alternating with more replete days interest me most. Human trials give us a reason to investigate the restriction side; the mouse studies give us a reason to investigate alternation.^{3 29}

In food terms, I would explore variation within an adequate diet: some days built more heavily around legumes, grains and other plant foods, with fish, dairy or other chosen animal foods on other days. Legumes’ relatively lower sulfur-amino-acid content makes them relevant here, although food choices alone do not guarantee a particular intake.²⁰

That is a candidate pattern, not a measured human optimum. I would start by looking at the actual amino-acid composition of meals before declaring a day “low methionine” because it happened to be vegan.

I would also keep adequate protein on rest days. Human exercise research shows enhanced muscle sensitivity to amino acids persisting into the following day. The gym may be closed; the building site is still operating.³⁴

Several days each month: the fasting-mimicking approach

#

There is human evidence for a periodic pattern. In a randomised trial, three monthly cycles of a five-day fasting-mimicking diet reduced body weight, body fat, blood pressure and IGF-1. Participants returned to their usual diet between cycles.³⁵

This is the closest of these examples to a studied human restriction-and-repletion cycle, but it changed calories, protein and other dietary components together. A later seven-day trial comparing low- and higher-protein fasting-mimicking diets found that both reduced weight and IGF-1. That makes calorie restriction a serious part of the explanation, rather than letting amino acids take all the credit.³⁶

I find periodic restriction scientifically interesting, but it would not be my default first experiment. It is a larger intervention, and the studies do not justify promising muscle preservation or rejuvenation from the schedule.^{35 36}

Entire low-amino-acid weeks?

#

This is where my enthusiasm drops. Longer continuous restriction can change human metabolism, as the trials above show. But I do not see a convincing reason to prefer alternating long depletion and repletion blocks over a simpler adequate diet with shorter variations. That is my judgment about the trade-off, not the result of a head-to-head trial.

The experiment I would most like to see would compare continuous moderate restriction with intermittent restriction at matched weekly calories, protein and target amino-acid intake, while measuring insulin sensitivity, muscle function and bone outcomes. Then we could ask whether the rhythm adds something beyond the weekly average. Until then, my ranking is: a sustainable daily rhythm first, thoughtful variation between days as the more interesting experimental option, and aggressive multi-day cycles further down the list.

My bet is still mostly plants

#

The food-level evidence gives me another reason to prefer that starting point. Large observational studies associate plant-protein substitution with lower mortality, especially when it replaces processed red meat. A meta-analysis also associated higher plant-protein intake with lower all-cause mortality, while animal protein overall did not show a significant association with all-cause mortality.^{37 38}

Food quality matters within that pattern. Plant-based diets emphasising whole grains, fruit, vegetables, nuts and legumes have different cardiovascular associations from patterns dominated by less nutritious plant foods. “Vegan” is a broad category. So is “stuff in a petrol station.”³⁹

There is trial evidence for a plant-rich mixed diet too. In the corrected PREDIMED analysis, Mediterranean diets supplemented with olive oil or nuts reduced major cardiovascular events in older adults at high risk. That gives a mixed diet a credible place in this discussion.⁴⁰

My working choice remains something like 70–80% plant-based, understood as a direction for assembling meals rather than a magic percentage of calories or protein. I would build it around beans, lentils, chickpeas, tofu, tempeh, whole grains, nuts and seeds, with deliberate protein sources. Fish and modest amounts of low-fat dairy can fit that pattern; red and processed meat deserve separate consideration.^{20 41}

I would check the total protein and indispensable amino acids, then look at how the mixture varies. I would rather do that than reduce total protein indiscriminately and hope that feeling weaker means I am ageing more slowly.

Fully vegan deserves its own article. B12, iodine, calcium, iron and vitamin D need attention; vegetarians also tend to have lower muscle creatine stores. One detail worth fixing now: vegan vitamin D does not have to be D2. Lichen-derived D3 exists, and D3 generally raises and maintains circulating vitamin D more effectively than D2.^{42 43 44}

I am convinced there is something useful here. The human interventions persuade me that amino-acid composition is a real lever for metabolic health; the animal work gives cycling a strong rationale for further study. My best bet for everyday eating is a nutritious, mostly plant-based diet that supplies enough protein, supports muscle, and leaves room for variation and regular breaks from eating.

The next question I want answered is how much the rhythm matters. That is an experiment I would like to read before someone turns it into a subscription.

Sources and study details

#

Study populations, interventions and endpoints live here for anyone who wants to follow the trail.