You are on page 1of 11

Reviews

Exercise adaptations: molecular


mechanisms and potential targets
for therapeutic benefit
Sean L. McGee   1 ✉ and Mark Hargreaves2 ✉
Abstract | Exercise is fundamental for good health, whereas physical inactivity underpins
many chronic diseases of modern society. It is well appreciated that regular exercise improves
metabolism and the metabolic phenotype in a number of tissues. The phenotypic alterations
observed in skeletal muscle are partly mediated by transcriptional responses that occur
following each individual bout of exercise. This adaptive response increases oxidative capacity
and influences the function of myokines and extracellular vesicles that signal to other tissues.
Our understanding of the epigenetic and transcriptional mechanisms that mediate the skeletal
muscle gene expression response to exercise as well as of their upstream signalling pathways
has advanced substantially in the past 10 years. With this knowledge also comes the opportunity
to design new therapeutic strategies based on the biology of exercise for a variety of chronic
conditions where regular exercise might be a challenge. This Review provides an overview
of the beneficial adaptive responses to exercise and details the molecular mechanisms involved.
The possibility of designing therapeutic interventions based on these molecular mechanisms is
addressed, using relevant examples that have exploited this approach.

Chronic diseases associated with inactivity and/or mechanisms involved suggests that repeated, transient
overnutrition are becoming increasingly prevalent in increases in the expression of exercise-​responsive genes
modern society. Exercise is known to have numerous in skeletal muscle confer exercise adaptations over
posi­tive health benefits and is efficacious in the pri- time and contribute to the positive effects of exercise2,4.
mary prevention of 35 chronic diseases, including Therefore, understanding the molecular mechanisms
type 2 diabetes mellitus, non-​alcoholic fatty liver disease, mediating this response might reveal new therapeutic
cardio­vascular disease and certain cancers1. Regular targets for the design of novel medicines to improve the
exercise results in a number of adaptive responses that treatment of a wide range of chronic conditions.
contribute to its positive health benefits2. Although many This Review provides a conceptual overview of
of these adaptations are observed primarily in skeletal exercise adaptations, how they impart a wide range
muscle, exercise also has positive effects in other tissues3. of health benefits and the hierarchy of the molecular
For various reasons, not all individuals are able to mechanisms involved. In addition, we highlight proof-
1
Metabolic Research Unit, meet the levels of exercise required to experience the of-​concept examples where an understanding of the
School of Medicine wide-​ranging health benefits that exercise confers. bio­logy of exercise has revealed molecular targets that
and Institute for Mental and Consequently, considerable interest has been shown by have been therapeutically exploited to deliver important
Physical Health and Clinical
Translation (iMPACT), Deakin
researchers in mimicking these effects with pharma- health benefits.
University, Geelong, Victoria, ceuticals. However, as exercise imparts such complex
Australia. perturbations to homeostasis depending on its mode, Skeletal muscle adaption to exercise
2
Department of Physiology, intensity and duration, it would be naive to envisage that Skeletal muscle is a key tissue for human locomotion
The University of Melbourne, the health benefits of exercise in their entirety could be and, by virtue of its mass and metabolic capacity, it also
Parkville, Victoria, Australia. mimicked with drugs. Nonetheless, knowledge of the has a major role in determining whole-​body functional
✉e-​mail: sean.mcgee@
biology of exercise could potentially be harnessed to capacity, exercise performance and health outcomes.
deakin.edu.au;
improve therapeutic approaches to combat chronic The contraction of skeletal muscle during exercise
m.hargreaves@
unimelb.edu.au diseases. Indeed, engaging only a fraction of the adap- increases force production and energy demand, with
https://doi.org/10.1038/ tive response to exercise might still confer important concomitant activation of intramuscular metabolic
s41574-020-0377-1 health benefits. Our understanding of the molecular pathways responsible for ATP generation as well as

NAture RevIewS | EnDoCrInology volume 16 | September 2020 | 495


Reviews

Key points mechanical force and stretch, metabolic perturbations,


disruptions to the acid–base balance and redox state,
• Exercise is effective in the primary prevention of 35 chronic diseases. elevated muscle temperature and increased circulating
• The adaptive response to exercise, which is mediated in part by transcriptional concentrations of adrenaline5. These upstream signals
alterations in metabolic and other genes, is an important contributor to these health can act by engaging transmembrane receptors and acti-
benefits. vating downstream signalling pathways, as is the case for
• Identifying the mechanisms that mediate the exercise adaptive response could hormonal signals such as catecholamines. Alternatively,
uncover molecular targets to guide the design of new medicines to better treat exercise-​associated signals can directly activate exercise-​
chronic diseases. responsive signalling molecules, such as AMP-​activated
• A number of signalling, epigenetic and transcriptional molecules identified as protein kinase (AMPK), to propagate the exercise signal.
contributing to the exercise adaptive response have been targeted pharmacologically Exercise induces alterations in cellular energy bal-
to deliver health benefits in proof-​of-​concept studies.
ance that can result in exercise intensity-​dependent
• A further understanding of the complexities of the molecular responses to exercise activation of AMPK in skeletal muscle12. Once acti-
will provide new opportunities to engage these mechanisms therapeutically.
vated, AMPK signals to downstream substrates (for
example, acetyl-​CoA carboxylase and the TBC1 domain
for the various physiological systems responsible for family member 4) that aim to restore energy balance13.
oxygen and substrate supply, carbon dioxide removal, Similarly, the calcium–calmodulin dependent protein
heat removal, fluid balance and overall homeostatic kinase II (CaMKII) is directly activated by increased
regulation5. Repeated bouts of exercise as part of a cytoplasmic calcium levels, such as those that occur dur-
training programme induce long-​lasting adaptations, ing muscle contraction14. Indeed, this kinase is activated
which are dependent on the maintenance of an ongoing by exercise in skeletal muscle15 and has been implicated
exercise stimulus. in various exercise responses, including gene expression
responses16 and glucose uptake17. It is important to note
Exercise modality. In exercise physiology, the classic that exercise provides pulsatile and transient signals that
distinction has been between training characterized by result in physiological adaptations, as opposed to persis-
low load, high repetition (endurance, for example, long tent signals that are often associated with pathological
duration cycling) and high load, low repetition (strength adaptations. For example, an increase in the levels of the
or resistance, for example, heavy weight lifting) exercise catecholamines adrenaline and noradrenaline during
bouts6,7. These are two extremes on a continuum of exercise has an important role in regulating substrate
muscle contractile activation, and training programmes metabolism and cardiovascular responses to exercise18.
often incorporate both at various times8. Since the early Catecholamines quickly return to pre-​exercise levels
2000s, a resurgence of interest has occurred in high 15–60 min after exercise, depending on its intensity
intensity interval training to induce cardiovascular and and duration19. However, persistently elevated levels of
metabolic adaptations9. catecholamines are associated with pathological cardiac
The primary adaptations to endurance training are hypertrophy, hypertension and other chronic health
increased cardiorespiratory fitness, as demonstrated conditions20. Similar transient signalling in response to
by increased maximal oxygen uptake and increased exercise is observed in skeletal muscle. The activity of
muscle oxidative capacity. This change is facilitated by AMPK in skeletal muscle returns to pre-​exercise levels
mitochondrial biogenesis and angiogenesis, with little, 3 h after the completion of exercise21, and chronic AMPK
if any, change in muscle mass. Despite it not increasing activity is associated with pathological adaptations such
muscle mass, endurance exercise can be effective in as cardiomyopathy22,23.
delaying and even preventing age-​related loss of muscle The main research focus on endurance training adap-
mass10. By contrast, strength training results in increased tations has been on the importance of calcium dynamics,
muscle fibre diameter and muscle mass; however, some metabolic perturbations, and reactive oxygen species
effects can be exerted on metabolic capacity. For example, production and their signalling via various kinases
endurance exercise results in increased mitochondrial (for example, CaMKII and AMPK) to the molecular
protein synthesis with no change in myofibrillar pro- processes involved in mitochondrial biogenesis6,7,24. By
tein synthesis, whereas resistance exercise is associated contrast, the research focus for resistance training has
with an increase in both myofibrillar and mitochondrial been on the various stimuli for muscle hypertrophy that
protein synthesis in untrained individuals11. This effect signal via the mechanistic target of rapamycin complex 1
might be partly related to the metabolic perturbations (mTORC1) pathway3,4. These signalling pathways are
that occur in both types of exercise, especially in the not independent and complex; indeed, convergent and
untrained state. divergent interactions probably occur depending on the
nature of the muscle contractile stimuli8. As an example,
Mechanisms of skeletal muscle exercise adaptation. the skeletal muscle expression of the glucose trans-
The traditional parameters of exercise training pro- porter GLUT4 is increased in response to endurance25,
grammes, such as load or intensity, duration and fre- sprint interval26 and resistance27 training. Currently,
quency, have multiple and diverse effects on the various it is unclear whether this effect reflects multiple signals
intra­muscular and hormonal signals that potentially converging on a common signalling network, a com-
mediate skeletal muscle adaptations. These signals mon signal targeting multiple signalling pathways, or
include but are not limited to increased catecholamine some combination of both. As is often the case in bio­
signalling, sarcoplasmic calcium release, alterations in logy, there are multiple, complex and redundant control

496 | September 2020 | volume 16 www.nature.com/nrendo


Reviews

systems that regulate homeostatic responses to stressors capacity in adipose tissue is a rodent-​specific pheno­
such as exercise. menon remains to be determined53 and could explain the
Studies of AMPK loss-​of-​function animal models discordance between rodent and human studies43,48–52.
have highlighted the complexity of exercise adaptive Cognitive function and brain health are also influ-
responses. For example, mice with deletion of a key enced by exercise in humans, and this is particularly
catalytic AMPK subunit or mice that overexpress a rele­vant to healthy ageing54. Emerging evidence also
kinase-​dead version of this subunit both show normal suggests that exercise influences the gut microbiome,
metabolic and adaptive responses to exercise 28–30. which has emerged as a major modulator of metabolism
In these AMPK-​defective mice, exercise activates protein in health and disease55.
kinase D, which has a similar substrate specificity to
AMPK but is not activated by exercise in wild-​type The role of exercise-​induced endocrine factors. Contr­
mice31; this finding emphasizes that redundant systems acting skeletal muscle is well known to release meta­
control important physiological responses such as bolites such as lactate, glycerol and alanine, which have
during exercise. Furthermore, studies utilizing phospho­ a key role in intermediary metabolism during exercise.
proteomics have observed that the phosphorylation of Over the past two decades, researchers have shown con-
many phosphosites was altered by various exercise siderable interest in other bioactive molecules released
regimes32–34, which raises the possibility that numerous from contracting skeletal muscle, the so-​called myo­
other exercise-​regulated kinases and substrates are kines. The interest in myokines was initially catalysed by
important in the response to exercise. observations related to exercise and muscle-​derived IL-6
In light of the key role that muscle oxidative capacity (ref.56). Numerous cytokines, peptides and meta­bolites
has in determining muscle metabolism, exercise are now recognized to be released from contracting
capacity and health35–37, the molecular mechanisms skeletal muscle during exercise57,58. These various fac-
responsible for the exercise-​induced increase in muscle tors could mediate many of the systemic benefits of
oxidative capacity are the focus of the current Review. exercise, for example, via effects on adipose tissue, liver57,
The regulation of muscle mass and exercise-​induced pancreas59, heart, blood vessels40 and brain60.
muscle hypertrophy are reviewed elsewhere38,39. An interesting question relates to the extent to
which muscle oxidative capacity might influence the
Effects of exercise on other tissues production and release of myokines before and after
Although much attention has focused on skeletal muscle exercise training. Of note, the release of bioactive mole­
adaptations to exercise, all organ systems of the body are cules during exercise is not limited to skeletal muscle.
affected by both acute bouts of exercise and long-​term Exercise-​induced adaptations in adipose tissue influ-
exercise programmes5,40 (Fig. 1a). The improved cardiac ence whole-​body glucose and lipid metabolism and
and vascular function with exercise training contri­ insulin action. In 2019, TGFβ2 was identified as an
butes to increased aerobic capacity and the protective exercise-​induced adipokine in mice that is sensitive
effects of exercise on cardiovascular disease risk40. The to lactate produced by contracting skeletal muscle61.
increased oxidative capacity specifically in the heart has Furthermore, TGFβ2 appeared to confer some of the
an important role in this response41. In skeletal muscle, beneficial effects of exercise training in glucose toler-
endurance training leads to substantial effects on ance61. Similarly, alterations in the circulating levels
energy metabolism35,36 as well as to effects in the liver42 of leptin seem to mediate exercise-​induced effects on
and adipose tissue43,44. For example, oxidative capacity haematopoietic progenitor cells and leukocyte pro-
is an independent predictor of the abundance of liver duction as well as an exercise-​associated reduction in
fat in humans45, which is important as the prevalence of cardiovascular inflammation in mice62. In addition,
non-​alcoholic fatty liver disease is increasing in parallel numerous exercise-​induced hepatokines exist in both
with that of obesity46. Furthermore, endurance exercise mice and humans that could mediate the beneficial
training reduces human liver fat even in the absence effects of exercise on metabolism, insulin action and
of weight loss, which is thought to be driven by mito- inflammation63. Finally, faecal microbiota transplanta-
chondrial biogenesis in hepatocytes that increases their tion from human marathon runners to mice improved
capacity to oxidize lipids47. exercise performance in the mice64, leading to the idea
Exercise training in rodents also increases mito­ that short-​chain fatty acids produced by gut microbiota
chondrial biogenesis and the oxidative capacity of white might mediate some of the benefits of exercise in various
adipose tissue48, whereas it is unclear whether these same organs55,64,65. Thus, the various peptides66, extracellular
metabolic alterations occur in response to exercise in vesicles67 and metabolites68,69 in plasma have key roles in
brown adipose tissue 43. In mice, these adaptations the inter-​organ communication that is likely to mediate
in white adipose tissue are associated with a beiging many of the effects of exercise. Supporting this concept,
phenotype, where white adipose tissue assumes some of we observed, in a 2019 study, that serum from volun-
the characteristics of brown adipose tissue, that increases teers who have undergone endurance exercise increases
fatty acid utilization and is thought to prevent the accu- GLUT4 expression in human adipocytes in vitro70 com-
mulation of ectopic lipid in other peripheral tissues43. pared with serum from unexercised volunteers. A key
However, a number of studies have found that exercise future challenge will be attaining a complete under-
training in humans had no effect on markers of oxidative standing of the full range of exercise-​induced factors,
capacity or beiging of subcutaneous adipose tissue49–52. their tissues of origin and their physiological relevance
Whether adipose beiging and increased oxidative to exercise adaptations.

NAture RevIewS | EnDoCrInology volume 16 | September 2020 | 497


Reviews

a
Gut
• Altered microbiome
Adipose tissue • Altered SCFA release Vascular system
• Enhanced substrate handling • Increased angiogenesis
• Altered adipokine secretion • Reduced atherosclerosis

Acute exercised and exercise-


Heart trained skeletal muscle
• Oxidative capacity Pancreas
• Altered morphology
• Myokines and extracellular • Enhanced β-cell function
• Improved cardiac function
vesicles
• Exercise-induced metabolites

Liver Brain
• Enhanced substrate handling • Improved cognition
• Enhanced insulin sensitivity Immune system • Enhanced mental health
• Altered hepatokine secretion • Reduced chronic
inflammatory state

b
Hierarchy of exercise Examples of exercise-sensitive
adaptive mechanisms potential therapeutic targets

Low High
• Catecholamines
Physiological stimulus
• Low energy balance

Potential therapeutic effect


On-target specificity
Intracellular signalling • AMPK
pathways • CaMKII

Epigenetic regulators Class IIa HDACs

• Nor1
Transcription factors
• PPARδ
and co-regulators
• MEF2 High Low

Exercise adaptive Enhanced oxidative


response capacity

Fig. 1 | Exercise adaptations and their theoretical therapeutic potential. a | Single bouts of exercise and exercise
training influences skeletal muscle oxidative capacity and induces the secretion of myokines, extracellular vesicles and
metabolites that facilitate crosstalk with other tissues. This crosstalk is thought to have positive effects on the function
of major systems and organs, including the brain, heart, liver, pancreas, adipose tissue, gut, and the immune and vascular
systems. b | Exercise imparts physiological stimuli or stressors that activate a complex array of intracellular signalling
pathways. Some of these pathways signal to epigenetic regulators, transcription factors and transcriptional co-​regulators
that precisely control transcriptional responses and the adaptive response to exercise. Therapeutically targeting
physiological stimuli that mediate the responses to exercise might induce a large proportion of the molecular responses
to exercise and have high theoretical therapeutic efficacy; however, such targets can have low on-​target therapeutic
specificity. That is, undesirable effects ensue when this target is engaged in the absence of the exercise stimulus in its
entirety. By contrast, targeting exercise-​associated transcriptional regulators has a fairly high on-​target therapeutic
specificity but might mediate only a small fraction of the exercise adaptive response and therefore have only low
therapeutic efficacy. AMPK, AMP-​activated protein kinase; CaMKII, calcium–calmodulin dependent protein kinase II;
HDACs, histone deacetylases; MEF2, myocyte enhancer factor 2; Nor1, nuclear receptor NR4A3; PPARδ, peroxisome
proliferator activated receptor-​δ; SCFA, short-​chain fatty acid.

Exercise-​induced transcription altered transcription and translation as well as in post-


As described earlier, exercise results in numerous adap- transcriptional and post-translational modifications7.
tations in multiple tissues and organ systems. A key A paradigm has emerged over the past 20 years that tran-
adaptation is the increased expression of vari­ous pro- sient increases in the levels of mRNA of vari­ous meta-
teins and changes in their activation state and locali­ bolic, myogenic and regulatory genes occur in response
zation as part of fundamental molecular and cellular to successive exercise bouts, which leads to increased
processes. The vari­ous signals generated during exercise protein expression and improved functional capacity2,4.
activate multiple signalling pathways that result in This paradigm was based on observations that a single

498 | September 2020 | volume 16 www.nature.com/nrendo


Reviews

bout of exercise increased transcription and the mRNA and was associated with their increased expression
levels of selected genes in skeletal muscle71–74. Subsequent in human skeletal muscle82. Of note, the removal of
studies in both rodents and humans have undertaken methyl groups from cytosine nucleotides requires
more detailed time-​course analyses of exercise-induced the catalytic action of ten-eleven translocation (TET)
changes in mRNA and protein levels as well as enzy- dioxygenase DNA demethylation enzymes83. Although
matic functional activity in skeletal muscle following decreased DNA methylation has been associated with
short-​term training over 1–2 months and confirmed increased expression of exercise-​responsive genes,
the importance of transcriptional events for adaptation mechanistic corroboration of the role of DNA hypo-
during and after exercise75,76. That said, these findings in methylation in the exercise-​induced transcriptional
no way diminish the role of post-transcriptional, trans- response remains to be established. Nonetheless, these
lational and post-​translational events or the effects on data provide the rationale for further investigations of
protein turnover and stability; these processes are just as the TET enzymes as mediators of the exercise-​induced
important, if not more so, in facilitating exercise adap- transcriptional response.
tations in skeletal muscle77. Furthermore, the accurate Post-​translational modifications to histone proteins
assessment of exercise adaptation requires more than just have a substantial influence on chromatin structure and
the measurement of mRNA levels78. Nevertheless, the gene expression, and have been extensively reviewed
study of transcriptional responses to exercise and elsewhere84–86. One such modification is acetylation.
the associated signalling and downstream molecular Exercise is associated with the acetylation of various
events does provide key insights on the molecular basis lysine residues within histone proteins in human skele­
of skeletal muscle adaptations to exercise. tal muscle87, a modification that is generally asso­ciated
with decompaction of chromatin and activation of
Epigenetic mechanisms of exercise adaptation. Alterations transcription88. Indeed, the acetylation of lysine 9 and 14
in gene expression require the regulation of epigenetic within the Glut4 promoter is associated with increased
machinery and the remodelling of chromatin (Box 1) expression of the Glut4 gene in rat skeletal muscle
to a transcriptionally permissive state. The epigenetic immediately following swimming exercise89. Of note, the
mechanisms currently known to be regulated by exercise acetylation state of histones is controlled by the reciprocal
include DNA methylation state and modifications to his- actions of histone acetyltransferases (HATs) and histone
tone proteins, and have been reviewed elsewhere79,80. The deacetylases (HDACs)90. Our own work, as well as
methylation of cytosine residues within DNA is generally that of others91,92,93,94, has established that the class IIa
associated with transcriptional repression. Methylation HDACs have a key role in the transcrip­tional response
mediates this repression either by impairing the DNA to exercise. For example, in human skele­tal muscle
binding ability of transcription factors with GC-​rich following 60 min of cycling exercise, the class IIa HDACs
consensus binding sequences or by altering the chroma- are phosphorylated and exported from the nucleus, which
tin architecture so that it favours a compacted structure is associated with histone acetylation and the expression
associated with transcriptional silencing81. of exercise-​responsive genes91,92. Furthermore, in mouse
DNA methylation was generally thought to be a skeletal muscle, transient expression of a HDAC5 mutant
fairly stable modification; however, global DNA methy­ that cannot be phosphorylated and exported from the
lation is rapidly decreased in human skeletal muscle nucleus prevents the expression of exercise-​responsive
within 45 min after exercise82. Furthermore, decreased gene Ppargc1a 93. Similarly, we have shown that
methylation of the promoter regions of certain exercise- overexpression of class IIa HDAC mutants that cannot
responsive genes that are important for oxidative recruit a functionally essential co-repressor complex
capacity (including Ppargc1a, PPARδ, TFAM and results in an exercise-like transcriptional response94. These
MEF2A) was observed immediately following exercise data suggest that the regulation of class IIs HDACs is an
essential component of the adaptive response to exercise.
The identity of the HAT enzymes responsible for his-
Box 1 | The importance of chromatin architecture
tone acetylation in response to exercise has been more
Eukaryote DNA is wrapped around a core of histone proteins, which together are elusive. Three major HAT families, the GNTA, MYST and
referred to as a nucleosome. Each nucleosome consists of 147 base pairs of DNA, p300/CBP families95, acetylate histones and non-​histone
wrapped ~1.65 times around an octamer of histone proteins148. These nucleosomes proteins. The activity and genomic localization of these
occur in repeating units to form chromatin and the larger chromosomes that help three HAT families are controlled by distinct signalling
compact and organize the vast amount of genetic material in eukaryote cells148. mechanisms96. Interestingly, loss-​of-​function mouse
Chromatin architecture has a key role in regulating gene transcription. Highly
models for p300 and GCN5, which represent major iso-
compacted chromatin is associated with transcriptional silencing as transcriptional
activators cannot gain access to the gene regulatory and coding regions of DNA88.
forms from two of the three HAT families, show no differ-
However, the decompaction of chromatin reveals these DNA regions to transcriptional ence in metabolic adaptions to exercise when compared
regulators and is associated with transcriptional activation149. A number of factors with wild-​type mice97,98. This finding could suggest that
influence the structure of chromatin, including the methylation of cytosine residues other HAT enzymes are more important or, alternatively,
in DNA and post-​translational modifications of histone proteins, which are together that extensive redundancy exists in HAT signalling.
referred to as epigenetic modifications150. These modifications can be stable or highly
dynamic to control both steady state151 and pulsatile transcriptional kinetics152, Exercise adaption mechanisms at the transcriptional
respectively. These mechanisms are highly coordinated by numerous epigenetic level. Specificity in the activation of transcriptional pro-
enzymes, transcription factors and upstream signalling pathways to provide precise grammes is mediated by DNA binding transcription
control over specific transcriptional programmes.
factors99. They play a key role in coordinating epigenetic

NAture RevIewS | EnDoCrInology volume 16 | September 2020 | 499


Reviews

Type II fibres
responses at distinct genomic loci, which is mediated to activate metabolic transcriptional programmes117.
Muscle fibres that are fast through sequence-​specific DNA binding preferences100. For example, during exercise, AMPK in the nucleus
twitch, are able to produce Transcription factors recruit and interact with both tran- can phosphorylate the class IIa HDACs, which results
large amounts of tension, scriptional activators and repressors and play a key part in their nuclear export, histone acetylation and activa-
have a fairly high reliance
on anaerobic ATP production
in linking distal gene enhancers with target genes by tion of MEF2-​dependent transcription118. AMPK also
and fatigue easily. regulating the higher-​order chromatin architecture101,102. phosphorylates histone 3 at serine 10, which is required
These factors are also essential for recruitment of the before it can be acetylated at lysine 9 (ref.119). Taken
transcriptional initiation complex, which includes RNA together, these findings suggest that there is a highly
polymerase II, to regions of decompacted chromatin101. coordinated control of transcriptional programmes by
Numerous transcription factors have been identi- the energy-​sensing kinase AMPK.
fied as important in coordinating aspects of the exercise CaMKII is also activated by exercise15 and selectively
adaptive response. For example, nuclear hormone recep- phosphorylates HDAC4, which results in nuclear export
tors are important regulators of muscle phenotype, and of both HDAC4 and HDAC5 owing to the hetero­
their role in the adaptive response to exercise has been dimerization of these two class IIa HDAC isoforms120.
extensively reviewed elsewhere103. The nuclear hormone Pharmacological inhibition of CaMKII is sufficient
family includes transcription factors that are activated to prevent exercise-​induced histone acetylation at the
by lipid ligands such as the peroxisome proliferator acti- Glut4 promoter in rats89. AMPK and CaMKII are just
vated receptors (PPARs). Of note, the PPARδ (otherwise two examples of how exercise engages signalling path-
known as PPARβ) and PPARγ isoforms are important ways through perturbations in intracellular energy
for oxidative metabolism adaptations historically asso- and calcium homeostasis to control gene expression
ciated with endurance exercise, whereas the PPARα iso- responses in skeletal muscle. The complexities in the
form seems to be important for metabolic adaptations signalling responses to exercise are yet to be completely
in type II fibres103. deciphered. However, as multiple omics approaches are
Meta-​analyses of transcriptomic datasets from becoming more accessible, it is anticipated that further
human and mouse studies of both aerobic and resistance signalling mechanisms linking exercise with transcrip-
exercise revealed the nuclear receptor NR4A3 (Nor1) tional control will emerge. These data will probably be a
to be an important regulator of exercise adaptations, rich resource for potential targets to recapitulate aspects
independent of exercise modality104,105. This finding is of the exercise adaptive response.
supported by studies in transgenic mice overexpressing
Nor1 in skeletal muscle. In the sedentary state, these Therapeutic targets revealed by exercise
mice display an exercise training phenotype that is A mechanistic understanding of the hierarchical mecha­
characteristic of mixed exercise modalities, including nisms controlling the adaptive response to exercise has
skeletal muscle hypertrophy, increased vascularisation started to yield potential targets for therapeutic inter-
and improved endurance exercise performance106–108. ventions that increase oxidative capacity and provide
The myocyte enhancer factor 2 (MEF2) family of other exercise-​like health benefits. An important consi­
transcription factors has also been implicated in reg- deration when selecting targets for such a strategy is
ulating metabolic responses as well as muscle struc- the relative trade-​off between potential therapeutic
ture and function responses to exercise in mice and efficacy and specificity, assuming systemic pharmaco-
humans92,94,109. Although this is not an exhaustive list logical administration (Fig. 1b). For example, targeting
of exercise-​responsive transcription factors, it is inter- a physiological stimulus, such as catecholamines, will
esting to note that exercise-​responsive PGC1α is a recapitulate a large proportion of the exercise adaptive
transcriptional co-​activator for all of the transcription response in skeletal muscle; however, such a strategy will
factors mentioned110–112. Similarly, class IIa HDACs are have other undesirable effects such as increased hepatic
the major regulators of MEF2 function by directly asso- glucose output, hypertension and pathological cardiac
ciating with and repressing this family of transcription hypertrophy. These ‘on-​target’ effects become undesira-
factors113. These specific examples highlight how epige- ble in the absence of other specific responses to exercise;
netic enzymes are recruited to distinct gene regions to for example, elevated skeletal muscle glucose uptake
control specific transcriptional programmes. that requires increased hepatic glucose output or the
As mentioned earlier, the activity of these epigenetic increased oxygen requirement of skeletal muscle induced
regulators and transcription factors is controlled by by exercise that manipulates cardiac hypertrophy signals
post-​translational modifications, such as phosphor­ into a physiological cardiac hypertrophy phenotype. By
ylation and acetylation, that are regulated by exercise- contrast, targeting specific transcriptional mechanisms
responsive signalling pathways. AMPK, which is might increase on-​target specificity but engage only a
activated by a low cellular energy balance, is one of the small fraction of the exercise adaptive response, which
most extensively studied signalling molecules in the con- reduces the potential therapeutic efficacy.
text of exercise. Moderate intensity endurance exercise At present, the highly complex physiological
activates skeletal muscle AMPK in humans12, which responses to exercise seem impossible to replicate in
regulates aspects of substrate metabolism114,115 and cal- their entirety by pharmacological means. Furthermore,
cium homeostasis34. In addition to these roles, AMPK it is unlikely that any one drug will be able to repli-
also translocates to the nucleus during exercise116 and cate the entire adaptive response to exercise owing to
signals through a number of epigenetic regulators, the trade-​off between specificity and efficacy (Fig. 1b).
transcription factors and transcriptional co-​regulators This concept will be further discussed in the following

500 | September 2020 | volume 16 www.nature.com/nrendo


Reviews

proof-​of-​concept examples, whereby molecules regu- obesity and, importantly, did not result in the accumu­
lated by exercise have been targeted pharmacologically lation of cardiac glycogen or cardiac hypertrophy125. This
for therapeutic gain. In addition, the pharmacokinetics study emphasizes the importance of designing therapies
of potential therapies and their ability to mimic the with pharmacokinetics similar to the pulsatile nature of
pulsatile nature of exercise-​induced signalling and exercise-​induced signalling.
transcriptional responses will also be addressed.
Targeting nuclear receptors. Focusing on potential
Targeting AMPK signalling. The involvement of AMPK exercise-​associated transcriptional targets rather than
in the transcriptional response to exercise as well as in on signalling pathways has yielded interesting observa-
substrate uptake and utilization after exercise makes it an tions. Nuclear receptors are obvious molecular targets
obvious target for exercise-​like therapeutic interventions. to replicate aspects of the exercise adaptive response,
Early studies utilized 5-​aminoimidazole-4-​carboxamide particularly given their ligand-​specific requirements
riboside (AICAR), a compound that activates AMPK for activation, meaning that these transcription factors
in a manner similar to AMP121. Daily subcutaneous are highly druggable with small molecules. The PPARδ
administration of AICAR to rats for 4 weeks increased agonist GW501516 has received considerable interest
the abundance of mitochondrial enzymes in type II mus- for its potential to activate exercise-​like transcriptional
cle fibres as well as other exercise-​responsive metabolic programmes. For example, chronic administration
proteins, such as GLUT4, in all muscle fibre types122. of GW501516 for 5 weeks in mice induced a skeletal
It was later revealed that this AICAR-​associated effect muscle transcriptional programme that showed an
was largely driven by activation of a transcriptional ~50% overlap with that of endurance exercise train-
programme encoding these mitochondrial enzymes ing, comprising genes involved in lipid and mitochon-
and GLUT4 (ref.123). Both of these studies highlighted drial metabolism123. Increasing the dose of GW501516
that, although AMPK activation replicated some of the (ref.128), combining GW501516 with exercise training
skeletal muscle adaptive responses to exercise, it did not or administering GW501516 with AICAR123 potenti-
replicate all of them. ated these effects and induced many of the hallmark
The development of highly specific AMPK agonists features of endurance exercise training. Furthermore,
has enabled more robust assessment of AMPK as a thera­ GW501516 had positive effects on glucose homeostasis
peutic target. In particular, studies of the small mole- and other metabolic parameters in mouse models of
cule MK-8722, which is a direct allosteric activator of obes­ity and the metabolic syndrome129–131, highlight­
AMPK124, have been particularly revealing. For example, ing  the potential for therapeutic utility. However,
acute administration of a single dose of MK-8722 to whether these effects are mediated by skeletal muscle
mice induces a transcriptional response that is highly or other tissues, such as liver, adipose tissue or intestine,
similar to the response to a single bout of exercise in a remains to be determined132,133.
range of tissues, including skeletal muscle, heart, liver GW501516 has also been touted as a potential therapy
and adipose tissue125. Furthermore, ontology analysis for Duchenne muscular dystrophy owing to its activation
showed that this transcriptional response was enriched of genes involved in both oxidative capacity and muscle
for genes involved in mitochondrial biogenesis, fatty acid structure and function134. However, this compound was
oxidation, glucose uptake and insulin action125. Chronic ultimately withdrawn from clinical development owing
administration of MK-8722 for 1 month in obese mice to its propensity to induce the formation of colorectal
had exercise-​like effects on glucose homeostasis124. adenomas, which has been ascribed to PPARδ function135.
Of note, MK-8722 has also had positive efficacy in a However, this is an area of controversy as GW501516 and
rat model of chronic kidney disease, which is possibly another structurally related synthetic PPARδ agonist,
related to these positive effects on glucose homestasis126. GW0742, are reported to have anticancer activity in naso-
However, chronic administration of the drug also pharyngeal and skin carcinoma cells, respectively136,137.
resulted in cardiac hypertrophy, which is an undesirable Whether the divergent effects on tumorigenesis can be
effect thought to be secondary to the accumulation of ascribed to cell type-​specific PPARδ-​dependent tran-
glycogen in the heart124, presumably as a result of unre- scriptional programmes or whether they are a result
strained glucose uptake. Similarly, mice with a mutation of more complex interactions with different transcrip-
in the γ subunit of AMPK resulting in cons­titutive activa- tional co-​repressors or co-​activators in different cell types
tion of AMPK, show a fatal hypertrophic cardiomyopa- remains to be determined. Nevertheless, these discrepant
thy22,23,127. This finding highlights the pitfalls of increasing findings highlight potential problems with undesirable
substrate uptake by tissues without a concomitant increase on-​target effects in tissues other than skeletal muscle
in energy expenditure and is an example of the trade-​off when compounds designed to replicate aspects of the
between therapeutic efficacy and on-​target specificity. skeletal muscle exercise adaptive response are adminis-
Attempts to address the trade-​off between effi- tered systemically. Indeed, consideration of the action of
cacy and specificity have been made by manipulating a molecule in other tissues should be prioritized when
the meta­bolic properties of MK-8722. For example, selecting targets for therapeutic intervention.
MK-8722 analogues were synthesised with pharma-
cokinetics that were more similar to the transient and Targeting HDAC complexes. Our own work has focused
pulsatile nature of the exercise response125. These on modulating the MEF2–class IIa HDAC pathway
shorter-​acting MK-8722 analogues retained the ability in order to induce phenotypic adaptations in skeletal
to reduce hyperglycaemia in various rodent models of muscle (Fig. 2). The class IIa HDACs are catalytically

NAture RevIewS | EnDoCrInology volume 16 | September 2020 | 501


Reviews

a Fig. 2 | The MEF2–class IIa HDAC axis as a target to


replicate aspects of the exercise adaptive response.
a | At rest, myocyte enhancer factor 2 (MEF2) activity is
repressed by the class IIa histone deacetylases (HDACs),
which recruit a co-​repressor complex containing SMRT,
HDAC3 NCor1 and HDAC3 that keeps the surrounding chromatin
in a condensed state. b | During exercise, AMP-​activated
NCor1 SMRT
protein kinase (AMPK) is activated and translocates to the
nucleus, phosphorylating the class IIa HDACs, resulting
HDAC4 HDAC5
in their nuclear export. This disrupts the class IIa HDAC
MEF2 co-​repressor complex, resulting in histone acetylation
(Ac) and MEF2-​dependent transcriptional activation.
Gene c | Pharmacological targeting of the MEF2-​class IIa HDAC
expression axis with Scriptaid also disrupts the class IIa HDAC
co-repressor complex, resulting in histone acetylation
and MEF2-​dependent transcriptional activation and an
exercise-​like transcriptional response.

Nucleus
inactive against acetyl-​lysine due to a single amino acid
substitution in their active site138, which means that
they are a poor target for traditional HDAC inhibi­
tors designed to inhibit HDAC activity. The ability
b of class IIa HDACs to repress MEF2-​dependent gene
expression requires recruitment of a larger co-​repressor
complex that includes NCor1, SMRT and HDAC3
P (ref. 139) (Fig. 2a) . Exercise disrupts the interaction of
this co-repressor complex with the class IIa HDACs
AMPK Exercise following their phosphorylation and nuclear export91,92
HDAC3
(Fig. 2b); we therefore focused on identifying molecules
NCor1 SMRT that had similar effects. Importantly, the interaction
P P P P P P P P between the class IIa HDACs and SMRT occurs through
HDAC4 HDAC5 HDAC4 HDAC5
regions surrounding the class IIa HDAC active site140.
This information was used to identify molecules that
Ac can bind this site and have large structural capping
Ac
groups that protrude out of the active site and might
Gene
expression disrupt the molecular interactions that confer SMRT
Ac MEF2
Ac binding. This work identified Scriptaid as a molecule
Pol II that disrupts the interaction between HDAC5 and
SMRT94 (Fig. 2c).
A single administration of Scriptaid to wild-​type mice
induced an exercise-​like transcriptional response that
was dependent on an intact class IIa HDAC co-repressor
complex94. Furthermore, 4 weeks of Scriptaid admini­
c stration induced many of the hallmark pheno­t ypic
alterations observed with exercise training, including
HDAC3
an increased skeletal muscle oxidative capa­city and
improved exercise performance. In addition, 4 weeks
NCor1 SMRT of Scriptaid administration also increased whole-​body
energy expenditure and lipid oxidation and decreased
fasting blood concentrations of glucose and lipid
levels94. Scriptaid also had therapeutic effects in obese
Scriptaid
HDAC4 HDAC5 mice, increasing insulin-​stimulated glucose uptake into
skeletal muscle and improving cardiac function141. The
Ac
latter effect was an important observation as currently
Ac no therapies exist that can treat obesity-​associated
Gene cardio­myopathy. This finding was somewhat surpris-
Ac MEF2 expression ing considering the putative role of the class IIa HDACs
Ac
Pol II
in suppressing pathological cardiac hypertrophy142–144.
However, studies suggesting that the class IIa HDACs
suppress hypertrophy have been performed in mouse
models where class IIa HDACs have been deleted from
birth and could suggest that this family of HDACs have
divergent functions in the developing heart versus

502 | September 2020 | volume 16 www.nature.com/nrendo


Reviews

the adult heart. Whether the efficacy of Scriptaid on Conclusions


cardiac function in obesity is due to its direct action Exercise has numerous health benefits that are due,
on the heart or is secondary to increased lipid disposal in part, to skeletal muscle adaptations that can influ-
and decreased circulating levels of lipid remains to ence metabolism and function in other tissues and at
be determined. a whole-​body level. The transcriptional response to a
In addition to its ability to disrupt class IIa HDAC single bout of exercise is important for the phenotypic
co-​repressor interactions, Scriptaid is also a selective alterations of skeletal muscle in response to exercise
HDAC inhibitor against HDAC6. Although our studies training, and studies have begun to unravel the complex
confirmed that Scriptaid exerts its metabolic transcrip­ interactions between physiological stimuli, intracellular
tional effects by acting through the class IIa HDACs94, signalling pathways, and epigenetic and transcriptional
HDAC6 is an important regulator of cytoskeletal and mechanisms that control these responses. We have high-
proteasome function145. Therefore, Scriptaid lacks the lighted some of these mechanisms and proof-​of-​concept
specificity to be used clinically. Nonetheless, these examples, where exercise-​responsive molecules have
studies provide another example where exploring the been targeted pharmacologically to induce exercise-​like
bio­logy of exercise can reveal new preclinical thera­ adaptive responses and some associated health benefits.
peutic approaches to combat chronic diseases. These These examples show that it might not ever be possible to
data also highlight the value in identifying the speci­ faithfully mimic all of the complex molecular, physiolo­
fic molecular interactions involved in the adaptive gical and health effects of exercise; however, understand-
response to exercise, which might provide greater ing and pharmacologically harnessing the molecular
thera­peutic specificity than simple agonist or antago- events engaged by exercise in skeletal muscle can still
nist approaches to modulate exercise-​responsive mole­ deliver important health outcomes. Our understanding
cules. Indeed, interest is growing in the activation of of the molecular events and factors that respond to exer-
these molecular mechanisms in order to engage adap- cise in skeletal muscle are still rudimentary. However,
tive responses that are induced by exercise146. That this means that there will be copious opportunities to
said, given the diversity and complexity of the adaptive further leverage this approach as our knowledge of the
responses to exercise, it is unlikely that a single agent or molec­ular responses to exercise increases to combat
even a combination (that is, a polypill) will fully recapit- the raft of chronic diseases currently challenging society.
ulate these responses and their health and performance
consequences147. Published online 6 July 2020

1. Booth, F. W., Roberts, C. K. & Laye, M. J. Lack multifunctional CaM kinase may underlie decoding 25. Kraniou, G. N., Cameron-​Smith, D. & Hargreaves, M.
of exercise is a major cause of chronic diseases. of calcium signals. Neuron 12, 943–956 (1994). Effect of short-​term training on GLUT-4 mRNA
Compr. Physiol. 2, 1143–1211 (2012). 15. Rose, A. J. & Hargreaves, M. Exercise increases and protein expression in human skeletal muscle.
2. Egan, B. & Zierath, J. R. Exercise metabolism and the Ca2+-calmodulin-​dependent protein kinase II activity Exp. Physiol. 89, 559–563 (2004).
molecular regulation of skeletal muscle adaptation. in human skeletal muscle. J. Physiol. 553, 303–309 26. Burgomaster, K. A. et al. Divergent response of
Cell Metab. 17, 162–184 (2013). (2003). metabolite transport proteins in human skeletal
3. Febbraio, M. A. Exercise metabolism in 2016: This was the first study to quantify CaMKII muscle after sprint interval training and detraining.
health benefits of exercise - more than meets activation in human skeletal muscle in response Am. J. Physiol. Regul. Integr. Comp. Physiol. 292,
the eye! Nat. Rev. Endocrinol. 13, 72–74 (2017). to exercise. R1970–R1976 (2007).
4. Williams, R. S. & Neufer, P. D. in Handbook of 16. Ojuka, E. O., Goyaram, V. & Smith, J. A. The role of 27. Stuart, C. A., Lee, M. L., South, M. A., Howell, M. E. A.
Physiology. Exercise: Regulation and Integration CaMKII in regulating GLUT4 expression in skeletal & Stone, M. H. Muscle hypertrophy in prediabetic men
of Multiple Systems (eds Rowell, L. B. & Shepherd, J. T.) muscle. Am. J. Physiol. Endocrinol. Metab. 303, after 16 wk of resistance training. J. Appl. Physiol.
(American Physiological Society, 1996). E322–E331 (2012). 123, 894–901 (2017).
5. Hawley, J. A., Hargreaves, M., Joyner, M. J. & 17. Richter, E. A. & Hargreaves, M. Exercise, GLUT4, 28. Jorgensen, S. B. et al. Role of AMPKα2 in basal,
Zierath, J. R. Integrative biology of exercise. Cell and skeletal muscle glucose uptake. Physiol. Rev. 93, training-, and AICAR-​induced GLUT4, hexokinase II,
159, 738–749 (2014). 993–1017 (2013). and mitochondrial protein expression in mouse
6. Baar, K. Training for endurance and strength: 18. Watt, M. J., Howlett, K. F., Febbraio, M. A., Spriet, L. L. muscle. Am. J. Physiol. Endocrinol. Metab. 292,
lessons from cell signaling. Med. Sci. Sports Exerc. & Hargreaves, M. Adrenaline increases skeletal muscle E331–E339 (2007).
38, 1939–1944 (2006). glycogenolysis, pyruvate dehydrogenase activation and 29. Jorgensen, S. B. et al. Effects of α-​AMPK knockout
7. Hoppeler, H., Baum, O., Lurman, G. & Mueller, M. carbohydrate oxidation during moderate exercise in on exercise-​induced gene activation in mouse skeletal
Molecular mechanisms of muscle plasticity with humans. J. Physiol. 534, 269–278 (2001). muscle. FASEB J. 19, 1146–1148 (2005).
exercise. Compr. Physiol. 1, 1383–1412 (2011). 19. Warren, J. B., Dalton, N., Turner, C., Clark, T. J. & 30. Maarbjerg, S. J. et al. Genetic impairment of AMPKα2
8. Coffey, V. G. & Hawley, J. A. Concurrent exercise Toseland, P. A. Adrenaline secretion during exercise. signaling does not reduce muscle glucose uptake
training: do opposites distract? J. Physiol. 595, Clin. Sci. 66, 87–90 (1984). during treadmill exercise in mice. Am. J. Physiol.
2883–2896 (2017). 20. Goldstein, D. S. Plasma catecholamines and essential Endocrinol. Metab. 297, E924–E934 (2009).
9. Gibala, M. J. & Little, J. P. Physiological basis of brief hypertension. An analytical review. Hypertension 5, 31. McGee, S. L. et al. Compensatory regulation of HDAC5
vigorous exercise to improve health. J. Physiol. 598, 86–99 (1983). in muscle maintains metabolic adaptive responses and
61–69 (2020). 21. Wojtaszewski, J. F., Nielsen, P., Hansen, B. F., metabolism in response to energetic stress. FASEB J.
10. Konopka, A. R. & Harber, M. P. Skeletal muscle Richter, E. A. & Kiens, B. Isoform-​specific and 28, 3384–3395 (2014).
hypertrophy after aerobic exercise training. exercise intensity-​dependent activation of 5’-AMP-​ 32. Hoffman, N. J. et al. Global phosphoproteomic
Exerc. Sport. Sci. Rev. 42, 53–61 (2014). activated protein kinase in human skeletal muscle. analysis of human skeletal muscle reveals a network
11. Wilkinson, S. B. et al. Differential effects of resistance J. Physiol. 528, 221–226 (2000). of exercise-​regulated kinases and AMPK substrates.
and endurance exercise in the fed state on signalling 22. Blair, E. et al. Mutations in the gamma(2) subunit Cell Metab. 22, 922–935 (2015).
molecule phosphorylation and protein synthesis in of AMP-​activated protein kinase cause familial This was one of the first studies to quantify the
human muscle. J. Physiol. 586, 3701–3717 (2008). hypertrophic cardiomyopathy: evidence for the central extensive phosphorylation events that occur in
12. Chen, Z. P. et al. Effect of exercise intensity on skeletal role of energy compromise in disease pathogenesis. human skeletal muscle in response to exercise.
muscle AMPK signaling in humans. Diabetes 52, Hum. Mol. Genet. 10, 1215–1220 (2001). 33. Potts, G. K. et al. A map of the phosphoproteomic
2205–2212 (2003). 23. Burwinkel, B. et al. Fatal congenital heart glycogenosis alterations that occur after a bout of maximal-​intensity
This study uncovered the exercise intensity-​ caused by a recurrent activating R531Q mutation in contractions. J. Physiol. 595, 5209–5226 (2017).
dependent activation of AMPK in human skeletal the gamma 2-subunit of AMP-​activated protein kinase 34. Nelson, M. E. et al. Phosphoproteomics reveals
muscle. (PRKAG2), not by phosphorylase kinase deficiency. conserved exercise-​stimulated signaling and AMPK
13. Hardie, D. G., Schaffer, B. E. & Brunet, A. AMPK: Am. J. Hum. Genet. 76, 1034–1049 (2005). regulation of store-​operated calcium entry. EMBO J.
an energy-​sensing pathway with multiple inputs 24. Hood, D. A., Memme, J. M., Oliveira, A. N. & Triolo, M. 38, e102578 (2019).
and outputs. Trends Cell Biol. 26, 190–201 (2016). Maintenance of skeletal muscle mitochondria in health, 35. Gollnick, P. D. & Saltin, B. Significance of skeletal
14. Hanson, P. I., Meyer, T., Stryer, L. & Schulman, H. exercise, and aging. Annu. Rev. Physiol. 81, 19–41 muscle oxidative enzyme enhancement with
Dual role of calmodulin in autophosphorylation of (2019). endurance training. Clin. Physiol. 2, 1–12 (1982).

NAture RevIewS | EnDoCrInology volume 16 | September 2020 | 503


Reviews

36. Holloszy, J. O. & Coyle, E. F. Adaptations of skeletal 61. Takahashi, H. et al. TGF-​β2 is an exercise-​induced 81. Bird, A. P. & Wolffe, A. P. Methylation-​induced
muscle to endurance exercise and their metabolic adipokine that regulates glucose and fatty acid repression–belts, braces, and chromatin. Cell 99,
consequences. J. Appl. Physiol. Respir. Environ. Exerc. metabolism. Nat. Metab. 1, 291–303 (2019). 451–454 (1999).
Physiol. 56, 831–838 (1984). 62. Frodermann, V. et al. Exercise reduces inflammatory 82. Barres, R. et al. Acute exercise remodels promoter
37. Hesselink, M. K., Schrauwen-​Hinderling, V. & cell production and cardiovascular inflammation methylation in human skeletal muscle. Cell Metab.
Schrauwen, P. Skeletal muscle mitochondria as a via instruction of hematopoietic progenitor cells. 15, 405–411 (2012).
target to prevent or treat type 2 diabetes mellitus. Nat. Med. 25, 1761–1771 (2019). The first study to show that methylation of DNA
Nat. Rev. Endocrinol. 12, 633–645 (2016). 63. Ennequin, G., Sirvent, P. & Whitham, M. Role of linked to exercise-​responsive genes is reduced in
38. McGlory, C., van Vliet, S., Stokes, T., Mittendorfer, B. exercise-​induced hepatokines in metabolic disorders. response to exercise.
& Phillips, S. M. The impact of exercise and nutrition Am. J. Physiol. Endocrinol. Metab. 317, E11–E24 83. Rasmussen, K. D. & Helin, K. Role of TET enzymes
on the regulation of skeletal muscle mass. J. Physiol. (2019). in DNA methylation, development, and cancer.
597, 1251–1258 (2019). 64. Scheiman, J. et al. Meta-​omics analysis of elite Genes Dev. 30, 733–750 (2016).
39. Wackerhage, H., Schoenfeld, B. J., Hamilton, D. L., athletes identifies a performance-​enhancing 84. Bannister, A. J. & Kouzarides, T. Regulation of
Lehti, M. & Hulmi, J. J. Stimuli and sensors that initiate microbe that functions via lactate metabolism. chromatin by histone modifications. Cell Res. 21,
skeletal muscle hypertrophy following resistance Nat. Med. 25, 1104–1109 (2019). 381–395 (2011).
exercise. J. Appl. Physiol. 126, 30–43 (2019). This paper uncovered a muscle–gut–muscle 85. Jenuwein, T. & Allis, C. D. Translating the histone
40. Fiuza-​Luces, C. et al. Exercise benefits in cardiovascular signalling axis that regulates adaptation to code. Science 293, 1074–1080 (2001).
disease: beyond attenuation of traditional risk factors. exercise and exercise performance. 86. Lee, J. S., Smith, E. & Shilatifard, A. The language
Nat. Rev. Cardiol. 15, 731–743 (2018). 65. Hawley, J. A. Microbiota and muscle highway – of histone crosstalk. Cell 142, 682–685 (2010).
41. Vettor, R. et al. Exercise training boosts eNOS-​ two-​way traffic. Nat. Rev. Endocrinol. 16, 71–72 87. Gorisch, S. M., Wachsmuth, M., Toth, K. F., Lichter, P.
dependent mitochondrial biogenesis in mouse (2020). & Rippe, K. Histone acetylation increases chromatin
heart: role in adaptation of glucose metabolism. 66. Parker, B. L. et al. Multiplexed temporal quantification accessibility. J. Cell Sci. 118, 5825–5834 (2005).
Am. J. Physiol. Endocrinol. Metab. 306, E519–E528 of the exercise-​regulated plasma peptidome. Mol. Cell 88. Venkatesh, S. & Workman, J. L. Histone exchange,
(2014). Proteom. 16, 2055–2068 (2017). chromatin structure and the regulation of transcription.
42. Brouwers, B., Hesselink, M. K., Schrauwen, P. & 67. Whitham, M. et al. Extracellular vesicles provide a Nat. Rev. Mol. Cell Biol. 16, 178–189 (2015).
Schrauwen-​Hinderling, V. B. Effects of exercise training means for tissue crosstalk during exercise. Cell Metab. 89. Smith, J. A., Kohn, T. A., Chetty, A. K. & Ojuka, E. O.
on intrahepatic lipid content in humans. Diabetologia 27, 237–251.e4 (2018). CaMK activation during exercise is required for
59, 2068–2079 (2016). This study quantified the extensive number histone hyperacetylation and MEF2A binding at
43. Stanford, K. I. & Goodyear, L. J. Exercise regulation of proteins released from skeletal muscle in the MEF2 site on the Glut4 gene. Am. J. Physiol.
of adipose tissue. Adipocyte 5, 153–162 (2016). extracellular vesicles during exercise and showed Endocrinol. Metab. 295, E698–E704 (2008).
44. Thompson, D., Karpe, F., Lafontan, M. & Frayn, K. that they signal to the liver. This study linked the activation of CaMKII during
Physical activity and exercise in the regulation of 68. Lewis, G. D. et al. Metabolic signatures of exercise in exercise to lysine acetylation of histone 3 at the
human adipose tissue physiology. Physiol. Rev. 92, human plasma. Sci. Transl. Med. 2, 33ra37 (2010). GLUT4 promoter and provided one of the first
157–191 (2012). This work defined the numerous metabolites examples of epigenetic control of exercise-​induced
45. Kantartzis, K. et al. High cardiorespiratory fitness is altered in the plasma in response to exercise and transcriptional responses.
an independent predictor of the reduction in liver fat showed that a number of these metabolites have 90. McKinsey, T. A., Zhang, C. L. & Olson, E. N. Control
during a lifestyle intervention in non-​alcoholic fatty transcriptional effects in skeletal muscle cells. of muscle development by dueling HATs and HDACs.
liver disease. Gut 58, 1281–1288 (2009). 69. Brennan, A. M. et al. Plasma metabolite profiles in Curr. Opin. Genet. Dev. 11, 497–504 (2001).
46. Younossi, Z. M., Marchesini, G., Pinto-​Cortez, H. response to chronic exercise. Med. Sci. Sports Exerc. 91. McGee, S. L., Fairlie, E., Garnham, A. P. &
& Petta, S. Epidemiology of nonalcoholic fatty liver 50, 1480–1486 (2018). Hargreaves, M. Exercise-​induced histone modifications
disease and nonalcoholic steatohepatitis: implications 70. Flores-​Opazo, M., Raajendiran, A., Watt, M. J. & in human skeletal muscle. J. Physiol. 587,
for liver transplantation. Transplantation 103, 22–27 Hargreaves, M. Exercise serum increases GLUT4 5951–5958 (2009).
(2019). in human adipocytes. Exp. Physiol. 104, 630–634 92. McGee, S. L. & Hargreaves, M. Exercise and myocyte
47. van der Windt, D. J., Sud, V., Zhang, H., Tsung, A. (2019). enhancer factor 2 regulation in human skeletal muscle.
& Huang, H. The effects of physical exercise on fatty 71. Neufer, P. D. & Dohm, G. L. Exercise induces a Diabetes 53, 1208–1214 (2004).
liver disease. Gene Expr. 18, 89–101 (2018). transient increase in transcription of the GLUT-4 93. Akimoto, T., Li, P. & Yan, Z. Functional interaction
48. Trevellin, E. et al. Exercise training induces gene in skeletal muscle. Am. J. Physiol. 265, of regulatory factors with the Pgc-1α promoter in
mitochondrial biogenesis and glucose uptake in C1597–C1603 (1993). response to exercise by in vivo imaging. Am. J.
subcutaneous adipose tissue through eNOS-​dependent 72. Kraniou, Y., Cameron-​Smith, D., Misso, M., Collier, G. Physiol. Cell Physiol. 295, C288–C292 (2008).
mechanisms. Diabetes 63, 2800–2811 (2014). & Hargreaves, M. Effects of exercise on GLUT-4 and 94. Gaur, V. et al. Disruption of the class IIa HDAC
49. Flores-​Opazo, M. et al. Exercise and GLUT4 in human glycogenin gene expression in human skeletal muscle. corepressor complex increases energy expenditure
subcutaneous adipose tissue. Physiol. Rep. 6, e13918 J. Appl. Physiol. 88, 794–796 (2000). and lipid oxidation. Cell Rep. 16, 2802–2810 (2016).
(2018). 73. Pilegaard, H., Ordway, G. A., Saltin, B. & Neufer, P. D. One of the first studies to show that manipulation
50. Tsiloulis, T. et al. No evidence of white adipocyte Transcriptional regulation of gene expression in of protein interactions similar to exercise could
browning after endurance exercise training in obese human skeletal muscle during recovery from exercise. induce exercise-​like transcriptional and metabolic
men. Int. J. Obes. 42, 721–727 (2018). Am. J. Physiol. Endocrinol. Metab. 279, E806–E814 effects.
51. Camera, D. M., Anderson, M. J., Hawley, J. A. & (2000). 95. Ali, I., Conrad, R. J., Verdin, E. & Ott, M. Lysine
Carey, A. L. Short-​term endurance training does not One of the first studies to characterize the acetylation goes global: from epigenetics to
alter the oxidative capacity of human subcutaneous widespread transcriptional response to exercise metabolism and therapeutics. Chem. Rev. 118,
adipose tissue. Eur. J. Appl. Physiol. 109, 307–316 in the immediate post-​exercise period. 1216–1252 (2018).
(2010). 74. Pilegaard, H., Saltin, B. & Neufer, P. D. Exercise 96. Howlett, K. F. & McGee, S. L. Epigenetic regulation
52. Larsen, S. et al. The effect of high-​intensity training induces transient transcriptional activation of the of skeletal muscle metabolism. Clin. Sci. 130,
on mitochondrial fat oxidation in skeletal muscle and PGC-1α gene in human skeletal muscle. J. Physiol. 1051–1063 (2016).
subcutaneous adipose tissue. Scand. J. Med. Sci. 546, 851–858 (2003). 97. Dent, J. R. et al. Muscle-​specific knockout of general
Sports 25, e59–e69 (2015). 75. Perry, C. G. et al. Repeated transient mRNA bursts control of amino acid synthesis 5 (GCN5) does
53. Herz, C. T. & Kiefer, F. W. Adipose tissue browning precede increases in transcriptional and mitochondrial not enhance basal or endurance exercise-​induced
in mice and humans. J. Endocrinol. 241, R97–R109 proteins during training in human skeletal muscle. mitochondrial adaptation. Mol. Metab. 6, 1574–1584
(2019). J. Physiol. 588, 4795–4810 (2010). (2017).
54. Tyndall, A. V. et al. Protective effects of exercise 76. Egan, B., O’Connor, P. L., Zierath, J. R. & 98. LaBarge, S. A. et al. p300 is not required for
on cognition and brain health in older adults. O’Gorman, D. J. Time course analysis reveals gene-​ metabolic adaptation to endurance exercise
Exerc. Sport. Sci. Rev. 46, 215–223 (2018). specific transcript and protein kinetics of adaptation training. FASEB J. 30, 1623–1633 (2016).
55. Mailing, L. J., Allen, J. M., Buford, T. W., Fields, C. J. to short-​term aerobic exercise training in human 99. Segal, E. & Widom, J. From DNA sequence to
& Woods, J. A. Exercise and the gut microbiome: skeletal muscle. PLoS One 8, e74098 (2013). transcriptional behaviour: a quantitative approach.
a review of the evidence, potential mechanisms, and This paper provided an extensive time course Nat. Rev. Genet. 10, 443–456 (2009).
implications for human health. Exerc. Sport. Sci. Rev. analysis of the transcriptional and protein 100. Wilson, S. & Filipp, F. V. A network of epigenomic
47, 75–85 (2019). expression responses to exercise training in and transcriptional cooperation encompassing an
56. Pedersen, B. K., Steensberg, A. & Schjerling, P. human skeletal muscle. epigenomic master regulator in cancer. NPJ Syst.
Muscle-​derived interleukin-6: possible biological 77. Robinson, M. M. et al. Enhanced protein translation Biol. Appl. 4, 24 (2018).
effects. J. Physiol. 536, 329–337 (2001). underlies improved metabolic and physical 101. Lemon, B. & Tjian, R. Orchestrated response: a
57. Pedersen, B. K. & Febbraio, M. A. Muscles, exercise adaptations to different exercise training modes symphony of transcription factors for gene control.
and obesity: skeletal muscle as a secretory organ. in young and old humans. Cell Metab. 25, Genes Dev. 14, 2551–2569 (2000).
Nat. Rev. Endocrinol. 8, 457–465 (2012). 581–592 (2017). 102. Palstra, R. J. & Grosveld, F. Transcription factor
58. Whitham, M. & Febbraio, M. A. The ever-​expanding 78. Miller, B. F., Konopka, A. R. & Hamilton, K. L. The binding at enhancers: shaping a genomic regulatory
myokinome: discovery challenges and therapeutic rigorous study of exercise adaptations: why mRNA landscape in flux. Front. Genet. 3, 195 (2012).
implications. Nat. Rev. Drug. Discov. 15, 719–729 might not be enough. J. Appl. Physiol. 121, 594–596 103. Kupr, B., Schnyder, S. & Handschin, C. Role of nuclear
(2016). (2016). receptors in exercise-​induced muscle adaptations.
59. Barlow, J. P. & Solomon, T. P. Do skeletal muscle-​ 79. McGee, S. L. & Hargreaves, M. Epigenetics and Cold Spring Harb. Perspect. Med. 7, a029835
secreted factors influence the function of pancreatic exercise. Trends Endocrinol. Metab. 30, 636–645 (2017).
β-​cells? Am. J. Physiol. Endocrinol. Metab. 314, (2019). 104. Reibe, S., Hjorth, M., Febbraio, M. A. & Whitham, M.
E297–E307 (2018). 80. McGee, S. L. & Walder, K. R. Exercise and the skeletal GeneXX: an online tool for the exploration of transcript
60. Pedersen, B. K. Physical activity and muscle-​brain muscle epigenome. Cold Spring Harb. Perspect. Med. changes in skeletal muscle associated with exercise.
crosstalk. Nat. Rev. Endocrinol. 15, 383–392 (2019). 7, a029876 (2017). Physiol. Genomics 50, 376–384 (2018).

504 | September 2020 | volume 16 www.nature.com/nrendo


Reviews

105. Pillon, N. J. et al. Transcriptomic profiling of 122. Winder, W. W. et al. Activation of AMP-​activated IIa histone deacetylases (HDACs) to the SMRT/NCoR
skeletal muscle adaptations to exercise and protein kinase increases mitochondrial enzymes in transcriptional repression complex. J. Biol. Chem. 290,
inactivity. Nat. Commun. 11, 470 (2020). skeletal muscle. J. Appl. Physiol. 88, 2219–2226 18237–18244 (2015).
An extensive meta-​analysis that has characterized (2000). 141. Gaur, V. et al. Scriptaid enhances skeletal muscle
the skeletal muscle transcriptional response to 123. Narkar, V. A. et al. AMPK and PPARδ agonists are insulin action and cardiac function in obese mice.
different modes of exercise. exercise mimetics. Cell 134, 405–415 (2008). Diabetes Obes. Metab. 19, 936–943 (2017).
106. Goode, J. M. et al. The nuclear receptor, Nor-1, 124. Myers, R. W. et al. Systemic pan-​AMPK activator 142. Backs, J., Song, K., Bezprozvannaya, S., Chang, S. &
induces the physiological responses associated with MK-8722 improves glucose homeostasis but induces Olson, E. N. CaM kinase II selectively signals to histone
exercise. Mol. Endocrinol. 30, 660–676 (2016). cardiac hypertrophy. Science 357, 507–511 (2017). deacetylase 4 during cardiomyocyte hypertrophy. J. Clin.
107. Pearen, M. A. et al. The nuclear receptor, Nor-1, 125. Muise, E. S. et al. Pharmacological AMPK activation Invest. 116, 1853–1864 (2006).
markedly increases type II oxidative muscle fibers and induces transcriptional responses congruent to 143. Zhang, C. L. et al. Class II histone deacetylases act as
resistance to fatigue. Mol. Endocrinol. 26, 372–384 exercise in skeletal and cardiac muscle, adipose signal-​responsive repressors of cardiac hypertrophy.
(2012). tissues and liver. PLoS One 14, e0211568 (2019). Cell 110, 479–488 (2002).
108. Pearen, M. A. et al. Transgenic muscle-​specific 126. Zhou, X. et al. PAN-​AMPK activation improves 144. Vega, R. B. et al. Protein kinases C and D mediate
Nor-1 expression regulates multiple pathways renal function in a rat model of progressive diabetic agonist-​dependent cardiac hypertrophy through
that effect adiposity, metabolism, and endurance. nephropathy. J. Pharmacol. Exp. Ther. 371, 45–55 nuclear export of histone deacetylase 5. Mol. Cell.
Mol. Endocrinol. 27, 1897–1917 (2013). (2019). Biol. 24, 8374–8385 (2004).
109. Potthoff, M. J. et al. Histone deacetylase degradation 127. Arad, M. et al. Constitutively active AMP kinase 145. Seidel, C., Schnekenburger, M., Dicato, M. &
and MEF2 activation promote the formation of slow-​ mutations cause glycogen storage disease mimicking Diederich, M. Histone deacetylase 6 in health
twitch myofibers. J. Clin. Invest. 117, 2459–2467 hypertrophic cardiomyopathy. J. Clin. Invest. 109, and disease. Epigenomics 7, 103–118 (2015).
(2007). 357–362 (2002). 146. Fan, W. & Evans, R. M. Exercise mimetics: impact on
110. Michael, L. F. et al. Restoration of insulin-​sensitive 128. Fan, W. et al. PPARδ promotes running endurance by health and performance. Cell Metab. 25, 242–247
glucose transporter (GLUT4) gene expression in preserving glucose. Cell Metab. 25, 1186–1193.e4 (2017).
muscle cells by the transcriptional coactivator PGC-1. (2017). 147. Hawley, J. A., Joyner, M. J. & Green, D. J. Mimicking
Proc. Natl Acad. Sci. USA 98, 3820–3825 (2001). 129. Tanaka, T. et al. Activation of peroxisome proliferator-​ exercise: what matters most and where to next?
111. Puigserver, P. et al. A cold-​inducible coactivator of activated receptor δ induces fatty acid β-​oxidation in J. Physiol. https://doi.org/10.1113/jp278761
nuclear receptors linked to adaptive thermogenesis. skeletal muscle and attenuates metabolic syndrome. (2019).
Cell 92, 829–839 (1998). Proc. Natl Acad. Sci. USA 100, 15924–15929 (2003). 148. Even-​Faitelson, L., Hassan-​Zadeh, V., Baghestani, Z.
112. Vega, R. B., Huss, J. M. & Kelly, D. P. The coactivator 130. Chen, W., Wang, L. L., Liu, H. Y., Long, L. & Li, S. & Bazett-​Jones, D. P. Coming to terms with chromatin
PGC-1 cooperates with peroxisome proliferator-​ Peroxisome proliferator-​activated receptor δ-​agonist, structure. Chromosoma 125, 95–110 (2016).
activated receptor α in transcriptional control of GW501516, ameliorates insulin resistance, improves 149. Dultz, E. et al. Quantitative imaging of chromatin
nuclear genes encoding mitochondrial fatty acid dyslipidaemia in monosodium L-​glutamate metabolic decompaction in living cells. Mol. Biol. Cell 29,
oxidation enzymes. Mol. Cell. Biol. 20, 1868–1876 syndrome mice. Basic Clin. Pharmacol. Toxicol. 103, 1763–1777 (2018).
(2000). 240–246 (2008). 150. Talbert, P. B., Meers, M. P. & Henikoff, S. Old cogs,
113. McKinsey, T. A., Zhang, C. L., Lu, J. & Olson, E. N. 131. Bernardo, B. L. et al. Postnatal PPARδ activation and new tricks: the evolution of gene expression in a
Signal-​dependent nuclear export of a histone myostatin inhibition exert distinct yet complimentary chromatin context. Nat. Rev. Genet. 20, 283–297
deacetylase regulates muscle differentiation. effects on the metabolic profile of obese insulin-​ (2019).
Nature 408, 106–111 (2000). resistant mice. PLoS One 5, e11307 (2010). 151. Li, E. & Zhang, Y. DNA methylation in mammals.
114. Merrill, G. F., Kurth, E. J., Hardie, D. G. & Winder, W. W. 132. Dimopoulos, N., Watson, M., Green, C. & Hundal, H. S. Cold Spring Harb. Perspect. Biol. 6, a019133 (2014).
AICA riboside increases AMP-​activated protein kinase, The PPARδ agonist, GW501516, promotes fatty acid 152. Nicolas, D., Zoller, B., Suter, D. M. & Naef, F. Modulation
fatty acid oxidation, and glucose uptake in rat muscle. oxidation but has no direct effect on glucose utilisation of transcriptional burst frequency by histone acetylation.
Am. J. Physiol. 273, E1107–E1112 (1997). or insulin sensitivity in rat L6 skeletal muscle cells. Proc. Natl Acad. Sci. USA 115, 7153–7158 (2018).
One of the first studies to show that pharmacological FEBS Lett. 581, 4743–4748 (2007).
activation of AMPK could induce metabolic effects 133. Doktorova, M. et al. Intestinal PPARδ protects Acknowledgements
similar to exercise. against diet-​induced obesity, insulin resistance The authors acknowledge the many researchers whose work they
115. Aschenbach, W. G. et al. Effect of AICAR treatment on and dyslipidemia. Sci. Rep. 7, 846 (2017). have not been able to cite in this Review. Original work by the
glycogen metabolism in skeletal muscle. Diabetes 51, 134. Miura, P. et al. Pharmacological activation of PPARβ/δ authors was supported by the Diabetes Australia Research Pro­
567–573 (2002). stimulates utrophin A expression in skeletal muscle gram and the National Health and Medical Research Council
116. McGee, S. L. et al. Exercise increases nuclear AMPKα2 fibers and restores sarcolemmal integrity in mature of Australia.
in human skeletal muscle. Diabetes 52, 926–928 mdx mice. Hum. Mol. Genet. 18, 4640–4649 (2009).
(2003). 135. Gupta, R. A. et al. Activation of nuclear hormone Author contributions
117. McGee, S. L. & Hargreaves, M. AMPK and receptor peroxisome proliferator-​activated receptor-​δ The authors contributed equally to all aspects of the article.
transcriptional regulation. Front. Biosci. 13, accelerates intestinal adenoma growth. Nat. Med. 10,
3022–3033 (2008). 245–247 (2004). Competing interests
118. McGee, S. L. et al. AMP-​activated protein kinase 136. Borland, M. G. et al. Inhibition of tumorigenesis by Both authors hold equity in Imitex Pty Ltd., a drug discovery
regulates GLUT4 transcription by phosphorylating peroxisome proliferator-​activated receptor (PPAR)- start-​up that was spun out from papers published by the
histone deacetylase 5. Diabetes 57, 860–867 dependent cell cycle blocks in human skin carcinoma authors that have identified the MEF2-​class IIa HDAC axis as
(2008). cells. Toxicology 404–405, 25–32 (2018). a druggable target to replicate aspects of the exercise adap-
119. Lo, W. S. et al. Phosphorylation of serine 10 in 137. Ji, Y., Li, H., Wang, F. & Gu, L. PPARβ/δ agonist tive response. This pathway is discussed in the current manu­
histone H3 is functionally linked in vitro and in vivo GW501516 inhibits tumorigenicity of undifferentiated script, along with many others that have been studied within
to Gcn5-mediated acetylation at lysine 14. Mol. Cell 5, nasopharyngeal carcinoma in C666-1 cells by the field.
917–926 (2000). promoting apoptosis. Front. Pharmacol. 9, 648
120. Backs, J., Backs, T., Bezprozvannaya, S., McKinsey, T. A. (2018). Peer review information
& Olson, E. N. Histone deacetylase 5 acquires calcium/ 138. Lahm, A. et al. Unraveling the hidden catalytic activity Nature Reviews Endocrinology thanks M.H. Laughlin and
calmodulin-​dependent kinase II responsiveness by of vertebrate class IIa histone deacetylases. Proc. Natl M. Tarnopolsky for their contribution to the peer review of
oligomerization with histone deacetylase 4. Mol. Cell. Acad. Sci. USA 104, 17335–17340 (2007). this work.
Biol. 28, 3437–3445 (2008). 139. Fischle, W. et al. Enzymatic activity associated with
121. Corton, J. M., Gillespie, J. G., Hawley, S. A. & class II HDACs is dependent on a multiprotein complex Publisher’s note
Hardie, D. G. 5-aminoimidazole-4-carboxamide containing HDAC3 and SMRT/N-​CoR. Mol. Cell 9, Springer Nature remains neutral with regard to jurisdictional
ribonucleoside. A specific method for activating 45–57 (2002). claims in published maps and institutional affiliations.
AMP-​activated protein kinase in intact cells? 140. Hudson, G. M., Watson, P. J., Fairall, L., Jamieson, A. G.
Eur. J. Biochem. 229, 558–565 (1995). & Schwabe, J. W. Insights into the recruitment of class © Springer Nature Limited 2020

NAture RevIewS | EnDoCrInology volume 16 | September 2020 | 505

You might also like