In the following sections, we discuss the acute (Fig. 1) and chronic (Fig. 2) effects of exercise on each of the peripheral organs involved in the regulation of whole-body insulin sensitivity and metabolic health, and the associated systemic health effects. Unless otherwise specified, only data from human studies are presented. A growing body of work implicates hundreds or even thousands of proteins secreted by the main peripheral tissues (i.e. myokines by skeletal muscle , hepatokines by the liver  and adipokines by adipose tissue ) involved in the regulation of energy homeostasis and whole-body insulin sensitivity. However, only a few have been characterised and had their biological function demonstrated. During exercise, some of these proteins, collectively named ‘exerkines’, are secreted, and create a complex inter-organ network that contributes to the systemic metabolic health effects of exercise . We also posit that ‘substrate flux’ between organs may work in concert with exerkines or independently to induce adaptations. Here we mention the molecular factors that have been recently identified and thought to potentially coordinate the inter-organ crosstalk in response to exercise (Fig. 3). Undoubtedly, numerous other signalling molecules will be identified in the years ahead, and future research will need to delineate the respective contribution of such molecules. In addition, because this area is still relatively new, controversies exist, and whether the findings from animals can be translated to humans is questionable.
Skeletal muscle adaptations
Skeletal muscle is the largest metabolic tissue in the human body and a critical site for glucose disposal both at rest and during exercise . During exercise, skeletal muscle uses both muscle glycogen stores and circulating plasma glucose as sources of fuel. Muscle contractions, even at low intensity and low volume , activate both oxidative and non-oxidative glucose disposal and glucose uptake via insulin-dependent and -independent mechanisms , optimising insulin action and both glucose oxidation and storage. Continuation of regular exercise further augments skeletal muscle oxidative capacity, mitochondrial biogenesis  and mitochondrial quality control mechanisms (mitophagy, fission, fusion), although this has been less well examined [31, 32]. Endurance training promotes the trafficking of dietary fatty acids away from storage and towards oxidation via an increased capacity for myocyte fatty acid transport paired with increased fatty acid oxidation in mitochondria in both normal weight and overweight adults . In addition, exercise training increases intramuscular triacylglycerol turnover and reduces lipid intermediate (diacylglycerols and ceramides) content . Importantly, increased glucose utilisation by muscle during exercise would lead to hypoglycaemia if it was not paired with a rapid upregulation of hepatic glucose output. Thus, exercise also drives critical acute and chronic adaptations to hepatic metabolism.
As previously reviewed [26, 35], the skeletal muscle secretory response, including a constellation of myokines, extracellular vesicles and their cargo, and metabolites, has recently been implicated in exercise-mediated multisystemic adaptations that improve metabolic health. They can act in a paracrine/autocrine or endocrine manner. Although numerous molecular analytes have been detected, the functions of only a few are well established. For example, levels of brain-derived neurotrophic factor (BDNF), proteins belonging to the natriuretic peptide family (e.g. B-type natriuretic peptide [BNP]), musclin, IL-15, IL-6, apelin, secreted protein acidic and rich in cysteine (SPARC), fibroblast growth factor 21 (FGF-21), decorin, myonectin and irisin have been shown to increase in the circulation in response to exercise and these molecules are likely to play a role in the control of muscle mass and skeletal muscle metabolism in an autocrine fashion, as reviewed elsewhere [36, 37]. The effect of some of these molecular factors have been studied in rodents or in in vitro studies but have not been (fully) confirmed in human studies. Circulating levels of IL-6, the first myokine to be discovered , rapidly increase in response to an acute bout of exercise . Its secretion seems to be influenced by the mechanical workload of the exercise , skeletal muscle glycogen content  and, potentially, glucose ingestion and/or plasma glucose levels . IL-6 may therefore act as an energy sensor. In humans, exercise-released IL-6 stimulates the mobilisation of intramuscular triacylglycerol , fatty acid oxidation , translocation of GLUT4 from the cytosol to the membrane  and improves skeletal muscle insulin sensitivity .
During short bouts of activity and high-intensity exercise, muscles rely predominantly on intramuscular stores of glucose and fat. However, when exercise is sustained, a larger supply of substrates from outside the muscle is required . The requirement of glucose uptake for the working muscle must be paired with increased rates of hepatic glucose output to maintain euglycaemia . Exercise therefore increases hepato-splanchnic glucose flux, an effect that is not seen by sampling blood glucose, but, rather, requires tracer methodology to measure glucose turnover/flux. Exercise first increases the mobilisation of hepatic glycogen (the biggest glycogen store in the body) into plasma, and this is followed by increased rates of gluconeogenesis during longer exercise bouts . To fuel these processes, exercise also increases the uptake of the gluconeogenic precursors (lactate, pyruvate, glycerol) . Exercise-induced changes in gluconeogenesis are dependent on the rise in glucagon and the drop in insulin that occur during exercise . During each bout, the exercise-induced decrease in insulin sensitises the liver to the effects of glucagon. Exercise training, in the absence of weight loss, improves the ability of insulin to suppress glucose production by the liver . As in skeletal muscle, exercise stimulates a reduction in lipogenic processes and a simultaneous increase in lipid oxidation [47,48,49], which is likely to underlie the effects of habitual exercise in the prevention of NAFLD and maintenance of reduced intrahepatic lipid storage .
The liver faces other challenges during exercise, such as recycling metabolites, clearing toxic compounds, and buffering the by-products of lipid oxidation released by the muscle during exercise (e.g. medium-chain acylcarnitines) [45, 46]. The liver also produces ketones, which fuel neuronal tissues during exercise if there has been a prolonged period since the last meal . Thus, via adaptive responses in glucose and fatty acid metabolism, a well-controlled crosstalk exists between the liver and muscles to exchange substrates and maintain metabolic homeostasis during exercise.
Changes in liver metabolism during exercise are regulated, at least in part, by exercise-released myokines. IL-6 enhances hepatic fat oxidation and glucose production during exercise . Myonectin also improves systemic lipid metabolism. It does so by increasing liver fatty acid uptake through upregulation of fatty acid transporter genes, at least in rodents . In addition, during and immediately after exercise, liver hepatokines are released, including FGF21, follastin and angiopoietin-like 4 (ANGPTL4), which are involved in the regulation of circulating triacylglycerol concentrations, skeletal muscle mass and strength, and metabolism in rodents and humans [45, 53].
Adipose tissue adaptations
Whole-body fat oxidation rates increase with prolonged exercise or physical activity, particularly in postabsorptive conditions. The energy demands of muscles and the liver are met by NEFA mobilisation from adipose tissues, the largest source of stored energy in the human body. Increased mobilisation of NEFA from adipose, paired with increased oxidation, permits sustained exercise by delaying hypoglycaemia . Moreover, increased NEFA oxidation in the liver during exercise is necessary to fuel the high energy costs of gluconeogenesis. Storage and mobilisation of NEFA are under the control of insulin (lipogenic and anti-lipolytic hormone) and catecholamines (epinephrine and norepinephrine, lipolytic hormones) and the atrial natriuretic peptide (ANP, lipolytic factor) . During exercise, levels of circulating catecholamines, via sympathetic nervous system activation, and ANP release by the heart are enhanced and the plasma insulin level is decreased . The combined action of these factors induces lipolysis. Even low-intensity exercise is sufficient to increase adipose tissue NEFA mobilisation . Exercise training improves the sensitivity of adrenergic receptors to catecholamines in adipose tissue , while also enhancing markers of mitochondrial biogenesis and function , blood flow and glucose uptake . However, the improvements in adipose tissue metabolism and in blood flow in response to acute exercise are less pronounced in overweight and obese adults than in their lean healthy counterparts . An acute bout of exercise also increases lipoprotein lipase (LPL) activity . Although this may appear counterintuitive given that LPL stimulates fat storage in adipose tissue, it makes sense given the effect of an increase in systemic and muscle LPL activity  is the promotion of fat uptake by muscle. Importantly, because this latter effect is long lasting (12–18 h) after a single bout of activity , prior exercise reduces the net delivery of dietary fat to adipose tissue (arterial triacylglycerol)  and, potentially, fat storage. In line with this, exercise training leads to a modest reduction in adiposity even in the absence of weight loss . Furthermore, it decreases two variables with negative metabolic health outcomes, namely, central adiposity (at least in men)  and fat cell size (at least when associated with energy deficit). Although some studies suggested a greater effect of exercise on visceral than subcutaneous adipose tissue mass, potentially because visceral adipose tissue mass is more responsive to adrenergic activation , meta-analysis and reviews failed to report a differential effect on fat depots . The existence of an independent effect of regular exercise on visceral adipose tissue is currently debated. Finally, exercise in rodents has been shown to lower adipose inflammation, which tracks with improved whole-body insulin sensitivity ; however, data in humans suggest that exercise-induced changes in adipose inflammation are minimal unless paired with caloric restriction-induced weight loss . In summary, adipose tissue adaptations to exercise together contribute to improved systemic metabolic homeostasis and improved exercise performance.
Muscle–adipose tissue crosstalk also exists. As reviewed elsewhere , skeletal muscle release of IL-6, FGF-21 and irisin increases in response to exercise and influences adipose tissue metabolism, oxidative capacity and glucose uptake. Following the depletion of glycogen stores, IL-6 is released by the contracting muscles during exercise. IL-6 may stimulate adipose tissue lipolysis and NEFA mobilisation during exercise, and plays a major role in the reduction of visceral adipose tissue in response to exercise training in humans . However, other in vivo and in vitro studies are not as conclusive, and the role of IL-6 in adipose tissue biology is still under investigation. Similarly, the effects of muscle-released FGF-21 and irisin on adipose tissue in humans are either still unknown or under debate. A novel exerkine produced by skeletal muscle contraction has recently been identified that targets human adipose tissue to promote lipolysis, namely, growth and differentiation factor 15 (GDF15) . In terms of adipose tissue, adipokines modulate inflammation, lipid and glucose metabolism, blood pressure and atherosclerosis . Via its effect on fat mass, exercise can indirectly modulate levels of leptin and adiponectin, the two most well-studied adipokines, which are positively and negatively associated with fat mass, respectively. Although leptin and adiponectin have been associated with insulin sensitivity, the specific effect of exercise training on leptin and adiponectin is unclear . Recently, TGF-β2 has been shown to be secreted from adipose tissue in response to exercise and play a role in glucose homeostasis in mice . Results still need to be confirmed in humans.
Pancreatic glucagon and insulin orchestrate the regulation of blood glucose by facilitating glucose disposal in insulin-sensitive tissues and hepatic gluconeogenesis. Insulin secretion is primarily adjusted according to the amount of glucose taken up by beta cells. Insulin secretion during a glucose challenge is dramatically altered by exercise undertaken during the previous days and hours in both healthy individuals and in those with insulin resistance and type 2 diabetes . Exercise cessation drives up insulin secretion, while one bout of exercise can lower insulin secretion, showing that insulin secretion is tightly regulated by exercise-driven pathways . The pancreas contributes to the capacity of acute exercise to increase hepatic glucose production by reducing insulin secretion (hepatic insulin clearance is also likely to be important) and increasing glucagon secretion . In adults with impaired glucose tolerance and type 2 diabetes, exercise improves peripheral sensitivity and pancreatic beta cell function (greater insulin secretion in response to circulating glucose) . The combination of enhanced insulin sensitivity and improved beta cell function is defined as the disposition index, which is the product of insulin sensitivity multiplied by the amount of insulin secreted in response to blood glucose . Unlike drug therapies for type 2 diabetes, which typically only influence one component of the index independently, exercise has the capacity to improve the disposition index by enhancing both components . Emerging evidence shows that factors secreted from contracting skeletal muscle boost beta cell insulin secretion and that the improvements in glucose homeostasis in type 2 diabetes patients in response to chronic aerobic exercise may be more closely related to improved beta cell function than insulin sensitivity . Although controversies still exist and mechanisms have not been fully elucidated as yet, crosstalk between skeletal muscle and pancreatic alpha and beta cells seems to exist via myokines. Two of the proteins that have been proposed to play a role in the muscle–pancreas crosstalk are muscle-released IL-6, because of its link with the protective effect of exercise against proinflammatory-induced beta cell loss , and the peroxisome proliferator-activated receptor γ coactivator 1α (PGC1α)-dependent myokine irisin (precursor protein fibronectin type III domain-containing protein 5 [FNDC5]), because of its protective effect against beta cell apoptosis induced by lipotoxic conditions . Other unknown exercise-induced myokines may also play a role in modulating beta cell function. Further work in this area could highlight novel therapeutic targets for type 2 diabetes.
Endothelium and cardiovascular system adaptations
The skeletal muscle microvascular ensures that delivery of oxygen and substrates (NEFA, triacylglycerols-rich lipoproteins and glucose) matches the metabolic demands of the muscle fibres (and other metabolic tissues) under resting conditions and during exercise. The microvasculature of human skeletal muscles has a complex 3D structure and is subject to a large number of complementary blood-flow regulation mechanisms, but the exact cascade of events and regulatory processes remain unknown, especially in humans [81, 82]. Because of accessibility to tissues and the capacity to perform ex vivo preparations, rodents have served as an important model to explore the regulation of endothelial function and blood flow in the control of skeletal muscle metabolism.
Under resting conditions, insulin increases microvascular perfusion through both vasodilatory and vasoconstrictory activities. On the one hand, insulin acts on terminal arterioles and increases nutritive blood flow to skeletal muscle , which can also result in increased blood flow in upstream conduit arteries. On the other hand, data from rodents shows that insulin-mediated endothelin-1 (ET-1)  increases the vasoconstriction of arterioles that control access to the nutritive capillary beds of muscle, which receive little or no blood flow in the basal state. ET-1 in skeletal muscle arterioles is increased in individuals with obesity or type 2 diabetes compared with healthy control individuals . In addition, there is evidence in humans  and rodents  that increasing insulin resistance is associated with reduced capillary density. Although the vascular effects of insulin seem to be markedly compromised in type 2 diabetes, exercise-mediated pathways are likely to be maintained .
During exercise, there are increases in cardiac output, via a rise in cardiac stroke and heart rate, and blood pressure. Acute exercise increases muscle insulin sensitivity by a coordinated increase in insulin-stimulated microvascular perfusion and molecular signalling that improves glucose delivery and increases muscle glucose uptake and disposal . This insulin-stimulated increase in microvascular perfusion is likely to be linked to the exercise-mediated increases in muscle membrane permeability to glucose and muscle blood flow . The increased haemodynamic forces, i.e. shear forces exerted by blood flow, are also translated by the glycocalyx layer (glycoproteins and proteoglycans) located on the luminal surface of endothelium into a vasodilatory response. This pulsatile flow-induced shear stress and release of nitric oxide induces a dynamic regulation of vascular tone via ET-1 synthesis/release , which could also potentiate the non-nutritive route of the blood flow. Vasodilation and additional microvascular units expand the endothelial surface area, thus enabling the delivery of nutrients to the muscle. Exercise training reduces resting blood pressure, heart rate and cardiac hypertrophy, improves the vasodilator response of the muscle microvasculature to insulin and exercise  and enlarges the microvascular network via angiogenesis and arteriogenesis . These adaptations have been linked to a variety of changes in tissue metabolism and signalling, including the production and release of nitric oxide and prostacyclin from the vascular endothelium . The finding that vascular endothelial growth factor B (VEGF-B) produced by skeletal muscle links endothelial NEFA uptake to the oxidative capacity of skeletal muscle by controlling the expression of fatty acid transporter proteins in the capillary endothelium represents a major recent discovery . Regulation of the expression of these proteins may prevent lipotoxic NEFA accumulation, the dominant cause of insulin resistance in muscle fibres. The effects of exercise on endothelial adaptations and the underlying mechanisms continue to be explored and much is yet to be learnt, especially related to how adaptations occur in those with obesity or type 2 diabetes vs a non-diseased state [93, 94]. The ability of exercise to influence endothelial transport of insulin and glucose  and to mediate enhanced insulin-stimulated blood flow via both the nutritive and non-nutritive routes  is likely to be particularly important for the prevention of type 2 diabetes. Finally, emerging data from rodents indicate that exercise also positively impacts endothelial function and vascular biology in adipose depots  and is likely to play a role in other key organs, such as the brain, liver and pancreas.