REF-1425 · Sharples & Turner, 2023 · American Journal of Physiology-Cell Physiology · Paper · PubMed

What is the epigenetic muscle memory idea?

That training leaves DNA-methylation marks that outlast a break and prime the next block. The review is its proposers' own; the human data are small single-group studies.

Main caveat: The proponents' review. It presents its authors' model, built mostly on their own human studies and with a "New & Noteworthy" claim to be "the most…
limited, mixed evidenceMeasured on non-athletes and recreational people

Sharples and Turner review the two proposed mechanisms by which skeletal muscle "remembers" earlier training, defining muscle memory as "the capacity of skeletal muscle to respond differently to environmental stimuli in an adaptive (positive) or maladaptive (negative) manner if the stimuli have been encountered previously".

They separate it from motor learning in the nervous system, while granting that part of the greater strength response to retraining comes from improved neural activation.

On the cellular (myonuclear) account they are even-handed: mice given testosterone grew 31% in muscle cross-section after 6 days of later overload, against a non-significant 6% in controls, but nuclei per fibre fell during the washout; rodent models mostly retain added nuclei, though nuclei gained with loaded wheel running were lost on detraining; a meta-analysis puts nuclear gain at about 23% in rodent loading and 9% in human training, with rodents losing 6.6% on detraining; in humans REF-1360 added no nuclei and an older-adult study found nuclei lost in detraining, so "there are very few human studies" and it is "difficult to firmly conclude whether a cellular muscle memory occurs in human skeletal muscle".

The epigenetic account is the authors' own: cultured muscle cells kept a DNA-methylation mark of an inflammatory exposure for 30 divisions; in young men, hypomethylation from a resistance-training block was partly kept through 7 weeks off and some genes responded more on retraining (REF-1426); similar results in older adults (Blocquiaux, an independent group), in mice, and in their then-unpublished interval-training study (later REF-1427); one gene from that work, UBR5, was later shown to regulate muscle mass in rodents.

They propose that the two mechanisms work together, discuss a "negative" memory of muscle wasting, and suggest that training designed for a lasting memory might one day reduce the training needed — a hypothesis.

Our reading: the clearest statement of the epigenetic-memory position by its originators, candid that the cellular question in humans is unresolved; its claim that retraining builds muscle "more quickly and to a greater extent" rests on a 1991 study and the authors' own uncontrolled one, and its statement that lean mass doubled on retraining compares a second gain that includes what was kept with a first gain that did not (REF-1426).

Protocol

Design: invited narrative review (AJP Cell Physiology), no search strategy, no appraisal, no pooled data. n = not applicable (a synthesis of cited cell, rodent and human studies).
Sections: what muscle memory is; the cellular account (myonuclear accretion, satellite cells and their role in hypertrophy, retention or loss of nuclei in rodents and humans, limits of cross-species comparison, the fate of nuclei in atrophy); the epigenetic account (DNA methylation, the human resistance-training studies, the role of UBR5, ageing, other kinds of exercise, mouse nucleus-type methylomes); how the two may integrate; negative memory and whether exercise can reset it; future directions.
Funding and interests: "Adam P. Sharples group are funded by The Research Council of Norway project number 314157"; "The authors declare no conflicts of interest". The authors are the originators of the epigenetic-memory model and authors of most of the human epigenetic studies the review presents (REF-1426; both authors are also on REF-1427).

Key Results

TopicWhat the review statesEvidence as the review cites it (second-hand)
Definitionmuscle "respond[s] differently ... if the stimuli have been encountered previously", positively or negativelythe authors' 2016 definition
Faster, larger regrowththe hypertrophic and functional response to resistance training occurs "more quickly and to a greater extent" on retraining, even after muscle mass has returned to baseline; part of the strength gain is neuralStaron 1991; REF-1426
Cellular memory, steroids (mice)31% gain in EDL cross-section after 6 days of overload in testosterone-treated mice vs a non-significant 6% in controls, 20% greater after 14 days; controversial because nuclei per fibre fell after the washoutEgner 2013 (Gundersen laboratory)
Cellular memory, rodent trainingadded nuclei retained after ladder climbing (rats; first block in juveniles) but lost after loaded wheel running (mice)several rodent studies
Meta-analysisnuclear gain about 23% in rodent loading and about 9% in human training; rodents lose 6.6% of nuclei on detrainingone meta-analysis
Cellular memory in humansREF-1360 added no nuclei; in older adults detraining reduced nuclei; "difficult to firmly conclude whether a cellular muscle memory occurs in human skeletal muscle"REF-1360; older-adult studies
Epigenetic memory, young menhypomethylation from 7 weeks of training partly kept through 7 weeks off; hypomethylated sites "more than doubled" on retraining, "associated with the same magnitude (a doubling) increase in lean mass"REF-1426 (the authors' study)
Epigenetic memory, replicationsmore hypomethylation after retraining also in young adults after training, 2 weeks of cast immobilisation and retraining, and in older adults; UBR5 and AMOTL1 profiles repeatedBlocquiaux (independent group)
UBR5hypomethylated and raised on retraining in humans; knockdown in mouse muscle caused atrophythe authors' and collaborators' rodent and cell work
Endurance"epigenetic muscle memory has only been observed after resistance exercise"; 3 months of aerobic training left no retained gene expression after 9 months off; unpublished HIIT data show 1,190 sites with retained hypomethylationLindholm 2016; Pilotto (unpublished, later REF-1427)
Integrationmyonuclei accrued in growth may carry their own methylation marks; satellite cells could pass marks to new nucleimouse nucleus-type methylomes (Wen 2021)
Open questions"It is not currently known how long cellular or epigenetic memory lasts"; larger human studies are needed—
Practicetraining designed for a longer memory might reduce the volume or intensity needed — stated as hypothetical, needing studiesthe authors' proposal

Mechanisms

Measured (in the studies reviewed): retention of added nuclei in most rodent overload and steroid models but not after loaded wheel running; myonuclear loss in older adults after detraining; DNA-methylation changes after resistance training that partly persist through weeks of detraining in small human cohorts; loss of muscle mass when UBR5 is knocked down in mice.
Authors' interpretation: muscle stores a memory of earlier training in two complementary forms — retained myonuclei and retained epigenetic marks, which may interact through satellite cells — and the memory primes a faster, larger response to retraining; a negative memory of wasting may also exist.
Inference: the step from "marks persist" to "regrowth is faster because of them" is not shown in humans here: the human studies of marks are single-group designs in which a second block always follows the first (REF-1426), and in the human strength trial with an untrained control limb the previously trained leg regained no faster (REF-1360). The authors' interval-training data later showed marks retained with no retained or enhanced fitness (REF-1427).
Inference: the lean-mass "doubling" compares the gain from the original baseline after retraining (12.4%) with the first gain (6.5%); allowing for what was kept, the second gain was about 10% from the group means (our arithmetic from REF-1426), larger but not double.

Practical Application

What is the epigenetic muscle memory idea? That training leaves DNA-methylation marks that outlast a break and prime the next block. The review is its proposers' own; the human data are small single-group studies.
Strength comes back partly through the nervous system: the authors themselves grant that part of the greater strength response to retraining is improved neural activation, and they keep motor learning separate from muscle memory (REF-1360 attributes kept strength to it).
Do not plan around a "memory": the authors' idea of training for a longer-lasting memory is a hypothesis they say needs studies, and how long any memory lasts is unknown by their own account; returns follow SYNTH-36 (the base's own practice, not from this review).
What it is good for as a source: the epigenetic-memory position in its originators' words, and a fair summary of the rodent and human myonuclear data with their limits. What it does not show: that retained marks make regrowth faster in humans, that any memory lasts beyond the weeks to months studied, or anything settled for endurance fitness.

Limitations

The proponents' review. It presents its authors' model, built mostly on their own human studies and with a "New & Noteworthy" claim to be "the most comprehensive review" of the field; no search or appraisal.
The evidence it rests on is small. Human epigenetic studies in small groups of untrained young or older people, in single-group designs; the functional claim (marks cause faster regrowth) is not tested in humans.
Overstated summary of [REF-1426]. The lean-mass "doubling" on retraining compares gains from different starting points (see Mechanisms).
Unpublished data. The interval-training results it reports were unpublished at the time; the published study (REF-1427) differs in detail (five genes rather than six) and found no physiological memory.
Source text. Read from the accepted manuscript, which "may contain minor differences" from the published version.
Funding and interests: Research Council of Norway (project 314157), as printed; no conflicts declared. No device maker, federation or data platform is named.
How this entry was checked and changed
Why this level

the rubric's B+ for an authoritative narrative review from the leading group on epigenetic muscle memory, held two steps lower: it argues its own group's position on a contested question; the human epigenetic studies it presents as the core evidence are mostly its own or its collaborators' (among them REF-1426: 7 men, one group, no control), so that evidence is a few small single-group studies of a molecular marker; and the review itself concludes that the cellular question in humans is open and that the duration of any memory is unknown. Read from the full text (the accepted manuscript in the Norwegian research archive; the publisher site sits behind a Cloudflare check and was not used)

Full text (checked 2026-10-29)
Skeletal muscle memory defined as the capacity of skeletal muscle to respond differently to environmental stimuli in an adaptive (positive) or maladaptive (negative) manner if the stimuli have been encountered previously (full text)
Mice that had received earlier testosterone showed a 31% increase in EDL cross-sectional area after 6 days of mechanical overload compared with a non-significant 6% increase in controls; overall growth after 14 days was 20% greater (full text)
Myonuclear accretion increased by approximately 23% after loading in rodents and about 9% after training in humans; rodents can lose 6.6% of their myonuclei during detraining (full text)
Very few human studies with few participants; difficult to firmly conclude whether a cellular muscle memory occurs in human skeletal muscle (full text)
After retraining the number of hypomethylated DMPs more than doubled compared to the earlier training period, associated with the same magnitude (a doubling) increase in lean mass (full text)
To date, epigenetic muscle memory has only been observed after resistance exercise; aerobic training could not demonstrate retained transcript expression levels during 9 months of detraining after 3 months of aerobic training (full text)
Unpublished observations: 1,190 CpG sites with a retained hypomethylated profile into detraining after high intensity training (full text)
It is not currently known how long cellular or epigenetic memory lasts (full text)
Adam P. Sharples group are funded by The Research Council of Norway project number 314157; the authors declare no conflicts of interest (full text)