How Autophagy Is Measured in Fasting Studies

In 1992 Yoshinori Ohsumi’s group published micrographs of yeast vacuoles filling with spherical bodies 400 to 900 nm across. The cells had been shifted from nutrient medium into medium lacking nitrogen. A few bodies appeared within an hour; after three hours they filled the vacuole almost completely. Inside them were ribosomes, endoplasmic reticulum, mitochondria and glycogen granules at roughly cytosolic density. Ohsumi named them autophagic bodies.

The bodies were visible only because the cells could not digest them. Ohsumi used mutants lacking proteinase A, proteinase B and carboxypeptidase Y, and then showed that wild-type cells do the same thing when the protease inhibitor PMSF is present. Remove the inhibitor and the bodies disappear. That experimental trick is the root of the difficulty in how autophagy is measured in fasting studies of humans three decades later: what you can see is not the pathway running, it is the pathway blocked.

What Ohsumi found in yeast between 1992 and 1993

Autophagy is the bulk delivery of cytoplasmic material to a degradative compartment – the vacuole in yeast, the lysosome in animal cells – inside a double-membrane vesicle called an autophagosome. The concept dates to the 1960s and to Christian de Duve, who discovered the lysosome and coined the term. What did not exist before Ohsumi was a genetic handle on it.

The 1992 paper established the assay; the 1993 follow-up used it. Tsukada and Ohsumi isolated a mutant, apg1, that failed to accumulate autophagic bodies despite normal vacuolar proteinases. That mutant also lost viability faster than wild-type cells during nitrogen starvation, giving them a cheap selection screen. Using loss of viability as a first filter they selected 75 further apg mutants. All of them, apg1 included, fell into 15 complementation groups: at least 15 genes were required for autophagy in yeast.

Those genes were later renamed ATG. Their products have counterparts in human cells, which is why a screen in baker’s yeast ended up describing a pathway in neurons. The Nobel Assembly at Karolinska Institutet awarded Ohsumi the 2016 Nobel Prize in Physiology or Medicine on 3 October 2016 for his discoveries of mechanisms for autophagy, and its press release identifies the early-1990s yeast experiments as the point at which the genes became identifiable.

How autophagy is measured in fasting studies of humans

The standard human readout is the lipidated form of MAP1LC3B, written LC3B-II. LC3B-I is cytosolic; on autophagosome formation it is conjugated to a membrane lipid and becomes LC3B-II, which sits in the autophagosome membrane and is degraded with the cargo. Investigators run a western blot on peripheral blood mononuclear cells (PBMCs) and quantify LC3B-II, usually as a ratio to LC3B-I or to a loading control.

A single blot yields a number that cannot be interpreted on its own. A high LC3B-II signal is compatible with two opposite states: many autophagosomes being made, or autophagosomes made at the usual rate and not cleared. The trick Ohsumi used on yeast is therefore reproduced deliberately in human samples. Blood is split, one part treated with a lysosomal inhibitor such as chloroquine, and the difference between the inhibited and uninhibited arms taken as flux – the LC3B-II that would otherwise have been destroyed in the interval. Both recent human fasting trials used that design.

The constraint that follows is unglamorous and decisive. The tissue has to be one that can be sampled repeatedly and treated ex vivo within a short window, and in practice that means blood. Liver, brain and heart – the tissues the ageing literature cares about – are not sampled in healthy volunteers.

Why a single LC3B-II band is not a measurement of flux

The field’s own reference document is blunt about this. The fourth edition of the guidelines for monitoring autophagy, coordinated by Daniel Klionsky and running to 382 pages in the journal Autophagy, states that there continues to be confusion regarding acceptable methods to evaluate autophagy, especially in multicellular eukaryotes, and that no individual assay is perfect for every situation, which is why multiple techniques are called for in each setting.

Two further cautions bear directly on how fasting papers should be read. Several core components of the machinery participate in both canonical and noncanonical autophagic processes, so genetic blockade is expected to target two or more ATG genes acting at distinct steps rather than one. And because many proteins involved in autophagy also regulate other cellular pathways, including apoptosis, not all of them can serve as specific markers of a genuine autophagic response. A protein whose level moves during fasting has not thereby reported on autophagy.

What the human fasting trials have and have not shown

Until recently the honest summary was that nutrient restriction stimulates autophagy in animals and that this had never been shown in humans. Three recent publications changed that in a narrow way.

Study	Design	Participants	Autophagy readout	Reported outcome
Bensalem et al., The Journal of Physiology, 2025	Randomised, 6 months; standard care vs calorie restriction vs intermittent fasting plus time-restricted eating	121 adults with obesity	Flux of LC3B-II in PBMCs in the context of whole blood	No significant between-group difference at 2 months; difference in change from baseline between the intermittent-fasting arm and standard care at 6 months, P = 0.04, post hoc
Espinoza et al., GeroScience, 2025	Pilot randomised trial, 8 days; two fasting-mimicking formulations vs control	30 healthy adults	LC3B-II/LC3B-I ratio in PBMCs treated ex vivo with chloroquine	Between-group differences in change from baseline to the end of the 6-day intervention for body weight, fasting glucose, beta-hydroxybutyrate, HOMA-IR and autophagic flux, p < 0.05; not significant across all time points
Masedunskas et al., Nutrients, 2024	Randomised crossover trial protocol; three-day water-only fast with or without glycogen-depleting exercise	24 planned	Autophagic flux in PBMCs, primary outcome	Protocol publication – design and endpoints only, no results reported

Read the qualifiers. The six-month result is a post hoc comparison at a single time point, and the authors note the effect may be driven partly by a tendency for autophagy to decrease in the control group rather than to rise under intermittent fasting. The eight-day trial enrolled 30 participants, was sponsored by the company selling the formulation tested and lists two of its employees among the authors, and its own abstract records that differences were not significant across all time points. The three-day water-only fast has a registered protocol and no findings in that publication; its authors describe a notable lack of human studies on the topic.

Both measured effects sit in circulating white blood cells. Whether the same thing happened in liver or brain is not addressed by these datasets, and the trials do not claim it.

What would strengthen the human evidence

The gap is analytical, not a matter of enthusiasm. Four things would narrow it, none exotic.

Prespecified flux endpoints. A post hoc comparison at one of several time points is hypothesis-generating. The Sydney crossover protocol registers autophagic flux as the primary outcome.
Cell-type resolution. PBMCs are a mixed population. Fasting alters circulating immune cell composition, so a shift in the average signal can reflect a shift in which cells are in the tube rather than a change inside any one of them.
Orthogonal markers. SQSTM1/p62 turnover and direct autophagosome counts answer different questions from an LC3B ratio.
Calibration. Western blot ratios are not standardised between laboratories and no reference material defines a unit of flux. Reporting assay specificity and performance alongside the clinical result, as the 2025 Journal of Physiology paper did, is the exception rather than the norm.

Ohsumi’s advantage was that yeast lets you delete a gene and then look down a microscope. Human work has neither move, which is why a pathway mapped genetically in 1993 still has a thin and contested human literature.

Frequently asked questions

What did Ohsumi discover about autophagy in yeast?

Ohsumi’s group showed in 1992 that yeast starved of nitrogen accumulate autophagic bodies in the vacuole, visible only when vacuolar proteases are absent or inhibited. In 1993 Tsukada and Ohsumi screened for mutants that failed to do this; the 76 mutants recovered fell into 15 complementation groups, the first genetic map of the pathway, later renamed the ATG genes.

How many ATG genes did Ohsumi identify?

The 1993 FEBS Letters paper reported that the autophagy-defective mutants fell into 15 complementation groups, implying at least 15 genes required for autophagy in Saccharomyces cerevisiae. They were designated APG1 to APG15 and later unified under the ATG nomenclature. The pathway as understood today involves more genes than that screen recovered.

What is the difference between autophagy and autophagic flux?

Autophagy is the process; autophagic flux is its rate. A count of autophagosomes, or a single LC3B-II band, measures a standing pool at one instant. That pool can grow because more autophagosomes are formed or because fewer are cleared. Flux estimates throughput by comparing matched samples with and without a block on lysosomal degradation.

Why is chloroquine used in autophagy flux assays?

Chloroquine raises lysosomal pH and impairs degradation of autophagosome contents, so LC3B-II accumulates instead of being destroyed. The difference in LC3B-II between treated and untreated portions of the same sample approximates how much was being degraded. It is the ex vivo equivalent of the protease-deficient yeast strains in which autophagic bodies first became visible.

Does fasting increase autophagy in humans?

The human evidence is thin and recent. A 2025 randomised trial in 121 adults with obesity reported a difference in LC3B-II flux in blood cells between an intermittent fasting arm and standard care at six months (P = 0.04, post hoc), and a 30-person pilot of a fasting-mimicking diet reported between-group differences at the end of a six-day intervention. Both were measured in circulating white blood cells only.

Can autophagy be measured in human liver or brain?

Not routinely. Flux assays need live cells that can be treated with a lysosomal inhibitor and sampled more than once, which in healthy volunteers means blood. Post-mortem or biopsy tissue gives a static snapshot rather than a rate. This is why human fasting data describe peripheral blood mononuclear cells rather than the organs the ageing literature is interested in.

Is autophagy a usable biomarker of ageing?

Not yet in any standardised sense. Western blot ratios are not calibrated across laboratories, no reference material defines a unit of flux, and the field’s own guidelines state that no single assay is adequate and that several autophagy proteins are not specific to the pathway. Comparing a flux value between two published studies is currently not meaningful.

References

Research use only. Nonsensia Lab supplies analytical reference standards for laboratory and research applications. This article is published for scientific and educational purposes. It is not medical advice, it does not describe any use in humans, and nothing in it should be read as a recommendation to administer any substance to a person or animal.

Filed under: Longevity Science

This article is part of our guide to The Biology of Ageing: Hallmarks, Chemistry and What the Evidence Supports.

Autophagy: The Recycling Pathway Behind Fasting Research