Disabled macroautophagy
Autophagy is the cell's recycling service: it wraps up worn-out proteins and broken parts of the cell, digests them and reuses the pieces. When it fails, animals develop brain and heart disease, and it is thought to slow with age. In people, exercise raises its markers in muscle, the evidence for fasting is thin, no supplement or drug has been shown to restore it, and nobody can yet measure whether it is working in your tissues.
Exercise regularly, and keep a sustainable eating pattern such as time-restricted eating, without counting on any particular fast length to switch autophagy on. Don't take spermidine, urolithin A or rapamycin for autophagy: none has been shown to switch it on or slow aging in people.
The cell's recycling system, and why it became a hallmark
Disabled macroautophagy is the fifth of the twelve hallmarks of aging, promoted to a hallmark of its own when the framework was expanded from nine to twelve in 2023.[1] Until then it was counted inside loss of proteostasis. It was split out because autophagy clears not just proteins but entire organelles and other large cargo.[2]
Macroautophagy (from here on just autophagy) works like this. Damaged organelles, misfolded protein clumps and even invading microbes are sealed inside a double-walled bubble, the autophagosome. It fuses with a lysosome, the cell's acid-filled digestive compartment, and the cargo is broken down into amino acids and fatty acids for reuse.[3] There are two other routes: microautophagy, in which the lysosome swallows cargo directly, and chaperone-mediated autophagy, which handles only individual proteins tagged for it. Macroautophagy is the high-capacity route and the main one for bulky clumps and whole organelles.[4]
More than thirty autophagy-related proteins run the process in sequence.[5] A few names come up again below. An enzyme complex called ULK1 starts it. A protein called LC3 coats the growing bubble and stays on the finished one, which is why it is the standard marker that autophagy is running.[6] Autophagy can also be selective: a receptor protein, p62, links tagged junk to the bubble. The best-studied selective form is mitophagy, the removal of damaged mitochondria, whose failure is central to mitochondrial dysfunction. And a master switch in the nucleus, TFEB (transcription factor EB), turns on the whole set of autophagy and lysosome genes.[7]
When the cell decides to recycle
Whether autophagy runs is set by the same nutrient-sensing network covered under deregulated nutrient sensing.[8]
mTORC1 is the brake. When nutrients are plentiful, the growth hub mTORC1 (mechanistic target of rapamycin complex 1) is active and holds ULK1 off and TFEB out of the nucleus. Fed cells don't recycle; they grow.
AMPK senses low energy. When energy runs low, AMP-activated protein kinase (AMPK) switches on and suppresses mTORC1, releasing the brake. Whether it also presses the accelerator directly is now disputed: in cell experiments, an acute energy shortage led AMPK to hold ULK1 back rather than switch it on, apparently to save energy while keeping the machinery intact.[9][10] A 2025 review describes AMPK as playing a dual role rather than acting as a simple trigger.[11] This low-energy signalling is part of why fasting and exercise can induce autophagy. Aging may blunt the sensor: in old rats, muscle AMPK at rest matched that of young rats, but its response to exercise or a drug activator was blunted.[12] Whether this slows autophagy in older people has not been measured.
SIRT1 links recycling to energy state. This enzyme depends on NAD⁺ (nicotinamide adenine dinucleotide), a coenzyme central to energy production. In cell and mouse studies it strips chemical tags off core autophagy proteins, and cells lacking it cannot fully switch autophagy on when starved.[13]
Does it really fail with age? In animals, yes; in people, unclear
The cost of failing autophagy falls hardest on cells that cannot divide to dilute their damage, such as neurons and heart-muscle cells.[14] In most animal studies autophagy declines with age and damaged material builds up. For the normal human brain, though, the evidence is only indirect, mostly from gene-expression studies.[15] One human signal comes from body fluids: levels of ULK1 were lower at older ages in the blood and spinal fluid of 75 cognitively healthy adults, and also reduced in 316 people with Alzheimer's. A fluid marker is not a direct measure of autophagy in the brain.[16]
The first direct measurements in living people's cells complicate the story. In a small study, flux, the actual rate of recycling, was higher in immune T cells from donors aged 67 to 93 than from those aged 23 to 35,[17] and it rose with age in blood cells from 114 adults at higher risk of type 2 diabetes;[18] in skin cells, though, it fell with age in men but not in women.[19] One reading is that aging cells push recycling harder to cope with more damage, so a higher number in an older person is not necessarily a good sign. These measurements, and their limits, are on Autophagic flux.
Where autophagy does fail, the weak point may be the final digestion step rather than the building of bubbles. In brain tissue from people with Alzheimer's, immature autophagosomes pile up in swollen, damaged nerve branches, which suggests they stall on the way to the lysosome.[20] In five mouse models of Alzheimer's, the digestive compartments inside neurons lose their acidity well before amyloid plaques appear, and the worst-hit neurons fill with undigested bubbles in a pattern also seen in human Alzheimer's brain.[21] The picture is a traffic jam, with autophagy switched on harder while digestion fails, rather than a shortage.[22] These bubbles are rare in brains without Alzheimer's changes, so whether the same holds in ordinary aging is not established.
Where stalled recycling does damage
The brain is the clearest case, and the evidence is from animals. Failing autophagy lets amyloid-beta and tau accumulate in Alzheimer's, alpha-synuclein in Parkinson's and mutant huntingtin in Huntington's.[23] The causal direction has been shown in mice: animals engineered to lose a core autophagy gene in forebrain neurons after birth develop age-dependent neurodegeneration with a build-up of phosphorylated tau, though not the tangles of human disease. Deleting tau or blocking its phosphorylation significantly reduced the damage.[24] Alzheimer's-linked molecules converge on the pathway too. Amyloid-beta and the APOE4 gene variant, the commonest genetic risk factor for late-onset Alzheimer's, both disturb autophagy in lab studies, and people with Alzheimer's who carry APOE4 show low levels of two autophagy markers.[25] Losing the presenilin enzymes, which are mutated in early-onset familial Alzheimer's, disrupts lysosome function in cell and mouse studies, though laboratories still disagree on why.[26][27]
Timing matters: in advanced disease, where the lysosome is already broken, making more autophagosomes may backfire and pile up undigested bubbles.[28] And so far no autophagy-boosting drug has helped patients. In a randomised, placebo-controlled trial in 161 adults with amyotrophic lateral sclerosis (ALS, motor neurone disease), weekly intravenous trehalose, a sugar chosen because it switches on autophagy in animals, did not clearly slow decline over 24 weeks, and serious side effects were more than twice as common as on placebo (16% versus 7%).[29] Getting drugs into the brain is a further hurdle: in a small pilot without a placebo group in 10 older adults with mild cognitive impairment or early dementia, rapamycin never reached detectable levels in the fluid around the brain, and markers of brain injury and tau rose rather than fell, though disease progression may explain the rise.[30]
Heart. Heart-muscle cells depend heavily on autophagy, and in animal studies it falls with age.[31] In mice, switching off a core autophagy gene in the adult heart leads to a thickened, weakened heart, but forcing autophagy too high under heavy pressure load also worsened damage, so balanced flux is the goal rather than as much as possible.[32] Human evidence is thin and hard to read. In hearts removed at transplant from people with severe heart-muscle disease, a small fraction of cells held autophagic build-ups that the authors read as cell death, and reviews read such human findings as more autophagy rather than less; it is unclear whether that is a failed rescue attempt or part of the damage.[33][34]
Liver. The liver uses autophagy to digest fat as well as protein. In mice, blocking it makes fat build up in the liver,[35] and in obese mice liver autophagy was sharply reduced; restoring a core autophagy gene improved the liver's response to insulin, so the vicious cycle is shown causally only in animals.[36] Liver biopsies from people with fatty liver disease show signs of stalled recycling, more marked in the inflamed form than in simple fatty liver.[37]
Muscle. Autophagy keeps the muscle's stem cells, the satellite cells, healthy. In mice, when it fails with age or is switched off genetically, these cells slip into senescence instead of repairing tissue, and restoring autophagy reversed this in very old mice; autophagy is also lower in satellite cells from very old people.[38] Whether this blunts the human response to exercise has not been tested.[39]
Inflammation and senescence: why more is not always better
Failing mitophagy can stoke inflammation, a link mapped mainly in cells and mice. Damaged mitochondria leak reactive oxygen species and their own DNA, which trigger the NLRP3 inflammasome, a molecular alarm that releases the inflammatory messengers interleukin-1β and interleukin-18.[40] Mouse immune cells lacking key autophagy genes carry more damaged mitochondria and release more interleukin-1β.[41] That makes failing autophagy a plausible route into inflammaging, not a demonstrated one.
The relationship with cellular senescence cuts both ways. Healthy autophagy suppresses the slide into senescence, but once a cell is senescent it can also help the cell make the inflammatory proteins it secretes, the senescence-associated secretory phenotype (SASP).[42] In cultured cells, blocking autophagy delayed senescence, including that secretion.[43] So the goal in age is restoring normal flux, not maximising autophagy everywhere.
Autophagy also supports the immune system. In mice, T cells lacking it fail to form long-lived memory cells.[44] In people the evidence is correlational: in older vaccine recipients, T cells with less autophagy responded more weakly.[45]
Exercise: the lever with the best human data
Exercise raises autophagy markers in human muscle, though the evidence is markers rather than a measured recycling rate. In muscle biopsies from 12 moderately trained young men, the marker LC3-II was about 1.5 to 2 times its starting level two hours after an hour of cycling, and a marker of mitophagy was also higher. After eight weeks of training, their resting muscle held more of the proteins that regulate autophagy, though not more autophagosomes.[46]
Whether autophagy is needed for the benefits of training has been tested only in mice: mice with one working copy of an autophagy gene gained less mitochondrial content, blood-vessel growth and endurance from wheel running.[47] In older people the picture is less simple: in a pilot of five adults aged 77 to 88, twelve weeks of mild exercise improved physical function while autophagic flux in their blood cells fell.[48] For how to train, see zone 2 and resistance training.
Fasting: strong in animals, thin in people
Fasting and calorie restriction are the most robust inducers of autophagy in animals. Lowering nutrients suppresses mTORC1 and activates AMPK and SIRT1, and the longevity benefits of calorie restriction across species are thought to work substantially through autophagy.[49] Fasting depletes a fuel molecule, acetyl-CoA, which switches off an enzyme that otherwise keeps autophagy proteins inactive.[50] It also raises the body's own spermidine, which fasting needs to switch on autophagy and extend life in model organisms; whether that matters in people is not yet shown.[51]
In people the evidence is thin. Studies of 24- to 72-hour fasts have found inconsistent changes in autophagy markers in muscle and blood, in 9 healthy adults a 24-hour water fast did not raise blood-cell flux, and how long a person must fast to switch autophagy on is unknown; the popular "24 to 72 hours" figure comes from mouse studies.[52] In 50 middle-aged women with obesity who fasted for 24 hours three days a week for eight weeks, autophagy markers in thigh muscle showed no sign of activation (one rose after a 24-hour fast, several fell with weight loss).[53] The best randomised result is modest: in 121 adults with obesity, six months of intermittent fasting plus time-restricted eating left blood-cell flux higher than standard care, but only in an exploratory analysis, with no significant rise within the fasting group itself and no effect from plain calorie restriction.[54]
Extreme restriction that tips into malnutrition can harm immune function,[55] so a sustainable pattern such as time-restricted eating is the sensible default, though it has not been shown to raise autophagy in people either.
Supplements and drugs: nothing shown in people
Caloric-restriction mimetics are compounds meant to reproduce these effects without eating less; the clinical detail is on geroprotectors.
- Spermidine has the best observational case: in one cohort of 829 adults in northern Italy, the top third of intake had about 39% lower death rates from any cause than the bottom third, a link that weakened once diet quality was accounted for.[56] But the trial the field wanted was null: 12 months of a low-dose spermidine-rich wheat-germ extract (about 0.9 mg a day) in 100 older adults with subjective cognitive decline improved neither memory nor the biomarkers.[57]
- Resveratrol is sold as a SIRT1 activator, but how it works is disputed: the early lab signal depended on an artificial fluorescent tag and vanished without it,[58] mouse work points to an indirect route,[59] and other researchers argue it acts directly with certain natural substrates.[60] Either way, very little of an oral dose reaches the blood.
- Metformin activates AMPK.
- Urolithin A, made by gut bacteria from compounds in pomegranates and walnuts, is sold as a mitophagy activator. Its human trials have mostly missed their main targets,[61][62] and none measured mitophagy itself.
Rapamycin directly blocks mTORC1 and is the most reproducible drug lifespan-extender in animals: it extends life in yeast, worms, flies and mice, where mid-life dosing delays several age-related diseases.[63] In healthy people it is unproven. The largest and longest trial, in adults aged 50 to 85 taking 5 or 10 mg a week for 48 weeks, missed its main goal, and its compounded pills reached only about a third of the blood level of standard rapamycin.[64] The often-quoted immune benefit, a roughly 20% better flu-vaccine response in older volunteers, came from a related drug, everolimus.[65] In 40 sedentary adults aged 65 to 85 on a home exercise programme, 6 mg once a week did not improve their gains over 13 weeks, may have slightly blunted them, and brought more adverse events in total.[66] No trial has shown that it extends life or delays age-related disease in people.[67]
Be cautious about long-term use: in transplant and epilepsy patients it brings impaired wound healing and more infections, and it can cause glucose intolerance.[68] The full side-effect and drug-interaction list is on the rapamycin page. It is investigational, not a routine intervention.
You can't yet check it in your own body
The usual readouts, LC3 and p62 levels in accessible cells, are snapshots, not rates: a high level can mean brisk recycling or a blockage where bubbles pile up undigested. The one practical rate measurement is a blood test, in which a sample is split, half is treated with a drug that stops lysosomes digesting, and the extra LC3 that builds up gives the flux.[69] It is now used as the main outcome in some randomised trials, but it captures only blood cells taken out of the body, and in aged mice blood-cell flux did not match the heart or brain.[70] For rapamycin there is also no standard marker of whether the drug is actually suppressing mTORC1 in a given person.[71]
So no intervention can yet be confirmed to be working in anyone's muscles or brain. Treat any product or protocol sold as "boosting autophagy" accordingly.
Further reading
- López-Otín C, et al. Hallmarks of aging: An expanding universe. Cell 2023.[72]
- Kaushik S, et al. Autophagy and the hallmarks of aging. Ageing Res Rev 2021.[73]
- Nieto-Torres JL, et al. Macroautophagy and aging: The impact of cellular recycling on health and longevity. Mol Aspects Med 2021.[74]
- Nixon RA. Autophagy-lysosomal-associated neuronal death in neurodegenerative disease. Acta Neuropathol 2024.[75]
- Lee JH, et al. Faulty autolysosome acidification in Alzheimer's disease mouse models induces autophagic build-up of Aβ in neurons, yielding senile plaques. Nat Neurosci 2022.[76]
- Inoue K, et al. Macroautophagy deficiency mediates age-dependent neurodegeneration through a phospho-tau pathway. Mol Neurodegener 2012.[77]
- Brandt N, et al. Exercise and exercise training-induced increase in autophagy markers in human skeletal muscle. Physiol Rep 2018.[78]
- Singh S, et al. Increasing autophagy with fasting and diet: separating fact from fiction. Biochem Soc Trans 2026.[79]
- Hofer SJ, et al. Spermidine is essential for fasting-mediated autophagy and longevity. Nat Cell Biol 2024.[80]
- Moel M, et al. Influence of rapamycin on safety and healthspan metrics after one year: PEARL trial results. Aging (Albany NY) 2025.[81]
- Hands JM, et al. What is the clinical evidence to support off-label rapamycin therapy in healthy adults? Aging (Albany NY) 2025.[82]
- HEALEY ALS Platform Trial. Safety and efficacy of trehalose in amyotrophic lateral sclerosis (HEALEY ALS Platform Trial): an adaptive, phase 2/3, double-blind, randomised, placebo-controlled trial. Lancet Neurol 2025.[83]