Exercise and inactivity as modifiers of β cell function and type 2 diabetes risk

Abstract

Exercise and regular physical activity are beneficial for the prevention and management of metabolic diseases such as obesity and type 2 diabetes, whereas exercise cessation, defined as deconditioning from regular exercise or physical activity that has lasted for a period of months to years, can lead to metabolic derangements that drive disease. Adaptations to the insulin-secreting pancreatic β-cells are an important benefit of exercise, whereas less is known about how exercise cessation affects these cells. Our aim is to review the impact that exercise and exercise cessation have on β-cell function, with a focus on the evidence from studies examining glucose-stimulated insulin secretion (GSIS) using gold-standard techniques. Potential mechanisms by which the β-cell adapts to exercise, including exerkine and incretin signaling, autonomic nervous system signaling, and changes in insulin clearance, will also be explored. We will highlight areas for future research.

GENERAL BACKGROUND

The prevalence of obesity and type 2 diabetes (T2D) have reached epidemic levels, with over 1.9 billion people worldwide classed as overweight or obese, whereas 537 million people are living with diabetes (1, 2). It is important, therefore, that the mechanisms involved in the development of these diseases and how they can be prevented are understood. Exercise can protect against many diseases including obesity and T2D (3–6), whereas the cessation of regular exercise has been shown to rapidly increase factors associated with metabolic dysfunction such as hyperinsulinemia, insulin resistance, and adiposity (7–12). One way in which exercise is believed to prevent metabolic disease is through the maintenance of pancreatic β-cell function and health. However, the mechanisms by which regular exercise adapts insulin secretion from the β-cell and has beneficial effects of β-cell proliferation and protection remain obscure (Fig. 1). In addition, little is known about how cessation of regular exercise impacts the β-cell and whether this is a key factor involved in the metabolic dysfunction that occurs after stopping exercise.

Figure 1.

Mechanisms exercise and inactivity may impact β-cell function. β-Cell adaptations to regular exercise and cessation of physical activity/exercise may be caused by altered 1) stimulation from exerkines released from tissues including skeletal muscle and liver, 2) stimulation from glucocorticoids, 3) stimulation from incretins, 4) β-cell AMPK activity, 5), stimulation from the autonomic nervous system, and 6) insulin clearance altering circulating insulin levels and β-cell insulin sensitivity. AMPK, AMP-activated protein kinase; InsR, insulin receptor.

EXERCISE, EXERCISE CESSATION, AND METABOLIC DISEASE

Exercise is defined as programmed bouts of physical movement that occur in a defined period of the day. In contrast, physical activity (PA) can be defined as movement that occurs throughout the day and can range from light to moderate, to vigorous intensity. Exercise can certainly include longer bouts of continual physical activity that occur in daily life (i.e., a 20-min continuous walk as a part of a job) making it difficult to always dissect these terms. In addition, there is a significant overlap between the physiological effects of bouts of moderate to vigorous PA and programmed exercise. However, this review will primarily focus on aerobic or endurance exercise per se as there is much less information available on the role of higher intensity anaerobic training and resistance exercise training and cessation from those modalities on β-cell function. Regular exercise is a critical regulator of metabolic function in virtually every tissue including muscle, liver, adipose, brain, neurons, heart, vasculature, and bones (13, 14), whereas inactivity (defined here as a lack of daily exercise and/or moderate-vigorous PA, as opposed to excessive sedentary behavior) has been linked to long-term pathologies in these tissues (5, 15). An alternative view is that regular exercise is not something that improves health, but is instead obligatory for normal health and function (5). Daily PA was likely required for survival and reproduction. Technological advances in the last century have diminished daily PA to low levels (∼5,000 steps/day) and increased sedentary (sitting) time (>8–12 h/day) (16) in combination with low levels of programmed exercise. Thus, our genetically determined metabolic pathways likely require daily exercise or moderate-vigorous PA for normal function but are now stuck in an inactive/sedentary environment compounded by hypercaloric diets (17).

Endurance/aerobic exercise is a potent physiological stimulus for enhancing liver and skeletal muscle glucose uptake and insulin action (18–20). Muscle contractions directly promote increased glucose uptake to fuel ATP needs during exercise, and replenish glycogen stores immediately after exercise (21). Dual stimulation of glucose uptake by contraction and insulin provides additive or synergistic effects on glucose transport in health and disease (22–25). The mechanistic interactions in muscle likely involve changes in energy flux (23), activation of contraction-mediated signaling kinases (AMPK and CaMK), and insulin signaling (e.g., Akt, Akt substrate of 160 kD [AS160] and TBC1 domain family member 1 [TBC1D1]) converging to enhance glucose transporter type 4 (GLUT4) translocation (26, 27). Insulin action in the liver is also highly regulated by exercise (28, 29). Hepatic insulin stimulation reduces hepatic glucose output via downregulation of gluconeogenesis, a signal that is impaired with insulin resistance, resulting in hyperglycemia (30). Acute exercise enhances insulin action in insulin-resistant liver in rodents (31, 32) and only 7 days of exercise improves hepatic insulin resistance in previously sedentary participants with type 2 diabetes (T2D) (33). Long-term endurance exercise also enhances the capacity of insulin to inhibit adipose lipolysis (as shown by reduced free fatty acid levels during postprandial conditions), further impacting tissue fuel selection and maintenance of metabolic flexibility (defined as appropriately using fuel sources that are available during postprandial vs. postabsorptive conditions or when dietary fat or carbohydrate intake is modified) which is conversely negatively impacted by inactivity (34). Thus, chronic exercise potently improves systemic insulin sensitivity and glycemic control in previously sedentary individuals through multiple mechanisms involving metabolic adaptations in skeletal muscle, liver, and adipose that have been reviewed elsewhere (35).

There is much less known about the mechanistic metabolic effects of transitioning from regular exercise to physical inactivity that can regularly occur on vacations, travel holidays, or with illness. Cessation of daily endurance exercise in endurance athletes for 10 days nearly doubled the insulin response to an oral glucose tolerance test (7), while one bout of exercise nearly normalized the insulin response suggesting restoration of insulin sensitivity. Similarly, independent studies showed that cessation of daily endurance exercise for only 2 days reduced insulin-stimulated glucose disposal measured during a hyperinsulinemic-euglycemic clamp to levels measured in sedentary controls (8, 36). Studies in highly physically active city dwellers (>10,000 steps/day) that reduced activity to <1,500 steps/day for 2 wk caused elevated insulin responses to an oral glucose tolerance test (OGTT) (10) and reduced peripheral insulin sensitivity and insulin signaling in skeletal muscle measured by a hyperinsulinemic-euglycemic clamp (37).

These observations have been modeled in rodents using the wheel lock technique. Rats given chronic access to voluntary wheel running (normally run 4–10 km/day) displayed enhanced insulin-stimulated glucose transport in epitrochlearis muscle 5 h after their last exercise bout (i.e., active controls), but levels were diminished to those found in sedentary rats if wheels were locked for only 2 days (9). Insulin signaling (insulin binding, phosphorylation of insulin receptor and Akt) and GLUT4 protein were also reduced after 2 days of wheel lock (9). These studies suggest that insulin sensitivity in muscle is primarily regulated by the “contractions of exercise/PA” performed in the hours/days before measurement as we reviewed in Ref. 19. We and others have shown that one bout of exercise, or only 7–10 days of exercise, can dramatically improve or restore insulin action and glycemia in insulin resistance and T2D without altering adiposity (23, 33, 38). In support of this concept, a cross-sectional study using hyperinsulinemic-euglycemic clamps in >300 individuals found the best predictor of insulin sensitivity to be activity levels (accelerometry) in the 7 days before the clamp, not age or obesity status (39).

Transitioning from regular exercise or PA to inactivity also causes rapid increases in adiposity in rodents and humans. The wheel lock studies in rats described earlier caused rapid increases in body fat percentage and epidydimal fat mass after only 3 and 7 days (40, 41), an affect that occurred even if elevated food intake during the first few days of inactivity was restricted (12). The rapid expansion of adiposity after wheel lock was associated with 15-fold higher triglyceride synthesis (palmitate incorporation) into adipose and increased cell volume compared with active rats (42). In addition, highly physically active city dwellers (>10,000 steps/day) that reduced activity to <1,500 steps/day for 2 wk had an ∼10% increase in visceral adiposity on a normal diet (10, 37) and a 45% increase in visceral fat and 20% increase in total fat mass on a hypercaloric diet (extra 1,500 kcal/day) (11). The human inactivity study also revealed that reduced insulin sensitivity and hyperinsulinemia occurred in unison with increased adiposity. We have also shown that following wheel lock, rats have reduced fat oxidation in muscle, whereas hepatic fatty acid oxidation increases (43). Elevated fatty acid oxidation in the liver is also found in high-fat diet (HFD) feeding and obesity as a compensatory response to hepatic lipid overload (44, 45), an adaptation that has been linked to hepatic oxidative stress and injury (46). Exercise cessation in hyperphagic rats also increases de novo lipogenesis in the liver with an increase in liver triglycerides apparent by 4 wk of inactivity (43, 47). These results suggest that inactivity induced a repartitioning of glucose and lipids from oxidation and storage in muscle to enhanced storage in adipose and the liver as we have reviewed previously (48). Bedrest studies have revealed a similar repartitioning of substrates (34). One hypothesis is that hyperinsulinemia, which occurs during any transition to inactivity (bed rest, exercise cessation, etc.), plays a critical role in repartitioning of substrates and downregulation of insulin action in skeletal muscle (49, 50). Limited knowledge exists, however, on the underlying mechanisms of how exercise and exercise cessation influence insulin secretion, and if changes in insulin secretion are associated with ensuing metabolic changes.

EXERCISE, EXERCISE CESSATION, AND INSULIN SECRETION

Insulin Secretion in Humans

Exercise has been proposed to affect insulin secretion from β-cells, indirectly and directly, over both short- and long-time scales. The majority of research suggests that exercise training decreases insulin secretion, which has been detected in numerous studies that have examined the insulin response during an OGTT (51–53). In addition, cessation of regular exercise increases the insulin response during an OGTT (7, 10, 37, 54). The most relevant studies examining the effect of exercise and inactivity on insulin secretion in vivo however, come from studies utilizing the hyperglycemic clamp, the gold standard technique for measuring insulin secretion.

In healthy participants with normal glucose tolerance, both aerobic and resistance training interventions lead to a decrease in glucose-stimulated insulin secretion (GSIS), as highlighted in Table 1 (55–57, 63, 64). In support of these findings, insulin secretion in response to both glucose and arginine is significantly lower in endurance-trained individuals when compared with sedentary controls (66–69, 72). Furthermore, cessation of regular exercise increases insulin secretion (66, 70, 71). Following 5 days of exercise cessation in men who had been exercising regularly (≥5 × 1 h/wk) for at least 3 years, insulin secretion during a hyperglycemic clamp was 68% higher at a glucose concentration of 7 mM, equal to that observed in untrained men (66). In the same study, insulin secretion was significantly higher following exercise cessation at a glucose concentration of both 11 mM and 20 mM (66). Similar results have been observed following 14 days of exercise cessation in people who regularly exercise (≥5 × 45 min/wk), with insulin secretion during a 10 mM hyperglycemic clamp being 49% and 72% higher during the early-phase (0–10 min) and late-phase (10–180 min), respectively (70). The increase in late-phase insulin secretion observed following 14 days of inactivity leads to insulin secretion levels matching those of untrained control subjects (70). In addition, a significantly higher insulin secretion is observed in response to both 11 and 20 mM glucose following 7 days of bed rest (71). These clamp results support the higher insulin responses to an OGTT reported in exercise cessation studies and provide clear evidence that regular exercise lowers insulin secretion in response to glucose. This is most likely important in preserving β-cell function and prolonging the time to β-cell exhaustion, which highlights the importance of regular exercise in reducing the risk for the development of T2D.

Table 1.

Insulin secretion measured by hyperglycemic clamp in exercise and inactivity

Study	Participant Health	Age, yr	Sex	Exercise/Inactivity	Insulin Secretion in Hyperglycemic Clamp
Exercise interventions
(55)	Healthy	65 ± 1	Both	Aerobic (80% HRmax) 4 × 60 min/week for 9 months	−15% 0 to 10 min (P = 0.06)
					−23% 15 to 180 min
(56)	Healthy	80 ± 1	Both	Aerobic (85% V̇o2peak) 3 × 60 min/wk for 9 mo	−30% 10 to 90 min
(57)	Healthy	59 ± 2	Male	Aerobic (80% HRmax) 3 × 40 min/wk for 9 mo	−14% 0 to 10 min
					−13% 20 to 120 min
(58)	Obese	33 ± 7	Female	Aerobic (65% V̇o2peak) 4 × 60 min/wk for 6 wk	−22% (ns) 0 to 10 min
					−21% 20 to 120 min
(59)	Obese	15 ± 2	Male	Aerobic (75% V̇o2peak) 3 × 60 min/week for 3 mo	No change
(60)	IGT or Mild T2D	51 ± 2	Both	Aerobic (65% V̇o2peak) 60 min/day for 10 days	−46% 0 to 10 min
					−33% 0 to 45 min
(61)	T2D	52 ± 3	Male	Aerobic (75% V̇o2peak) 5 × 40 min/wk for 3 mo	Increased at 25 mM in moderate secretors, no change in low secretors.
(62)	T2D	60 ± 2	Both	Walking (55% V̇o2peak) 5 × 60 min/wk for 4 mo	No change
(62)	T2D	58 ± 2	Both	Intermittent walking (3 min cycles at 70% and 40% V̇o2peak) 5 × 60 min/wk for 4 mo	No change
(63)	Healthy	58 ± 2	Female	Resistance, 3 × 14 exercises/wk for 4 mo	−16% 100 to 120 min
(64)	Healthy	57 ± 2	Female	Resistance, 3 × 14 exercises/wk for 4 mo	−16% 100 to 120 min
(65)	Post-stroke 50% IGT	65 ± 2	Both	Resistance, 3 × 3 exercises/wk for 3 mo	−24% 0 to 10 min
					−26% 10 to 120 min
(59)	Obese	15 ± 2	Male	Resistance, 3 × 10 exercises/wk for 3 mo	No change
Trained vs. untrained
(66)	Healthy	25 ± 1	Male	Endurance trained (5 × 60 min/wk for >3 yr) vs. untrained	Lower in trained at 7 (−36%), 11 (−55%) and 20 (−75%) mM
(67)	Healthy	27 ± 1	Male	Endurance trained (5 × 45 min/wk) vs. untrained	Lower in trained from 0 to 10 min (−65%) and 10 to 45 min (−66%)
(68)	Healthy	21 ± 1	Female	Endurance trained (12 h/wk) vs. untrained	Lower in trained from 0 to 10 min (−48%), 10 to 60 min (−22%, ns)
(68)	Healthy	45 ± 1	Female	Endurance trained (12 h/wk) vs. untrained	Lower in trained from 0 to 10 min (−41%) and 10 to 60 min (-51%)
(69)	Healthy	55 ± 10	Male	Endurance trained (25–70 miles/wk running) vs. untrained	Lower in trained from 0 to 10 min at 7.5 (−44%, ns), 10 (−18%, ns), and 15 (−62%) mM and from 10 to 120 min at 7.5 (−33%, ns), 10 (−48%) and 15 (−59%) mM
Exercise cessation
(70)	Healthy	31 ± 1	Both	14 days of inactivity in trained (5–7 × 45 min/wk)	+49% from 0 to 10 min and +72% 10 to 180 min post inactivity
(66)	Healthy	24 ± 1	Male	5 days of inactivity in trained (5 × 60 min/week for >3 yr)	+68% at 7 mM, +17% at 11 mM and +25% at 20 mM post inactivity
(71)	Healthy	25 ± 1	Male	7 days of bed rest	Higher at 11 mM and 20 mM post bed rest

IGT, impaired glucose tolerance; T2D, type 2 diabetes.

A limited number of studies have examined the impact of exercise on insulin secretion in obesity and T2D using the hyperglycemic clamp. In comparison with healthy adults, aerobic training programs in adults with obesity, impaired glucose tolerance (IGT), and mild T2D decrease insulin secretion (58, 60). In addition, in a study of participants who had previously suffered a stroke, of whom 50% had IGT, a decrease in insulin secretion was observed following a resistance training program (65). Similar results have not been detected in adolescents, however, with aerobic or resistance training in boys leading to no significant difference in insulin secretion compared with a sedentary control group (59). Interestingly, a study by Dela et al. (61) found that an aerobic exercise program for 3 mo led to a significant increase in insulin secretion at 25 mM glucose in people with T2D. This only occurred, however, in participants classified to have a “moderate” retained level of insulin secretion and not a “low” level of insulin secretion (plasma C-peptide greater than or less than 1.1 nmol/L 6 min following 1 mg of intravenous glucagon, respectively). This finding highlights the potential for exercise to induce context-dependent changes in insulin secretion. An exercise-induced increase in β-cell function in patients with T2D is supported by studies examining the disposition index (DI), a measure of β-cell function calculated as the product of GSIS and insulin sensitivity (73). As highlighted by Solomon (73), however, the hemoglobin A1c (HbA1c) status and length of time with T2D before an exercise intervention in human subjects appear to play a critical role in the effectiveness of exercise to improve glycemia; effects that seem to be primarily linked to improvements in the DI. Studies of DI, therefore, also support the idea that exercise-induced improvements in insulin secretion may only occur in people with a certain level of retained insulin secretion. Improved β-cell function as measured by DI has been observed in prediabetic people with overweight, obesity, and impaired fasting glucose (74, 75). No change in insulin secretion is observed in people with T2D following 4 mo of walking programs (62), suggesting a more vigorous or longer physical activity program may be required to induce beneficial effects on insulin secretion.

Insulin Secretion in Rodents

Like in humans, a decrease in insulin secretion in response to glucose has been observed in vivo using an OGTT (76) or hyperglycemic clamp (77). The ability to isolate pancreatic islets from rodents and expose them to controlled physiological conditions provides an opportunity to directly test the effect exercise training has on insulin secretion from β-cells and investigate potential mechanisms by which this is altered. Several studies utilizing male Wistar rats that performed aerobic exercise training (>3 times/wk) for 8 to 10 wk have reported a decrease in GSIS from isolated islets at both low (2.8–5.6 mM) and high (16.7–20 mM) glucose concentrations (78–86), with similar results observed in Sprague-Dawley rats (77, 87). Similar findings have been observed in rodent models of metabolic dysfunction, with a decrease in GSIS observed from isolated islets at high glucose (16.7 mM) from monosodium glutamate (MSG)-Wistar rats (78, 82, 88) and HFD Wistar rats at both high (16.7 mM) and low (5.6 mM) glucose (89). Other studies, however, have reported GSIS from isolated islets from male Wistar rats exposed to similar exercise programs to increase (90, 91) or not be significantly different (89). In addition, an increase in GSIS from isolated islets has been reported in control and MSG-Swiss mice following aerobic exercise training (92) and in C57BL/6 mice following resistance exercise training (93). These differing results may be due to methodological differences, such as the exercise protocol, timing of euthanasia, and the islet isolation and perifusion protocol, or the genetic differences between rodent models used in the studies. Most of the research in this area, however, gives further support for the role of exercise to adapt β-cells to secrete less insulin given the same glucose exposure. Minimal research has been conducted in female rodents, with only one study being conducted finding no difference in insulin secretion in female Sprague-Dawley rats following an aerobic exercise program (94). In addition, insulin secretion from rodent islets following the cessation of regular exercise has not been studied, meaning the mechanisms behind the rapid increase in insulin secretion seen in humans have not been investigated.

Studies of isolated islets from rodent models of T2D, with a reduced capacity to secrete insulin, contribute to the concept that exercise training may cause a beneficial, context-dependent increase in insulin secretion in situations of reduced β-cell function. Voluntary wheel running in Zucker diabetic fatty (ZDF) rats prevented a decrease in insulin secretion both during an OGTT and in isolated islets stimulated with glucose or glucose + palmitate, compared to inactive ZDF rats (95, 96). In addition, insulin secretion in 90% pancreatectomized rats following 10 wk of treadmill running is significantly higher than in sedentary rats. (97). In further support of this, insulin secretion was found to be significantly higher in 90% pancreatectomized rats when using a hyperglycemic clamp or in situ pancreatic perfusion (98). In addition, exercise training induced an increase in insulin secretion from islets isolated from streptozotocin-treated A/J mice at a low glucose concentration, with a nonsignificant increase observed at a high glucose concentration (99). These findings suggest that rather than causing a decrease in insulin secretion, as observed under healthy conditions, exercise may improve insulin secretion to improve glucose tolerance even in extreme experimental conditions.

Prevention of Hyperinsulinemia to Avert Metabolic Disease

As highlighted earlier, studies in both humans and rodents indicate that regular exercise training decreases insulin secretion in response to both high and low concentrations of glucose in most situations. In addition, cessation of regular exercise can lead to a rapid increase in insulin secretion. The simultaneous increase of insulin resistance, hyperinsulinemia, and adiposity in the development of metabolic dysfunction has led many to question which of these is the main driver (100, 101). Hyperinsulinemia has often been considered to occur in response to insulin resistance; however, many studies have challenged this dogma and suggest that hyperinsulinemia can play a causal role in the development of insulin resistance and obesity (100, 101). Hyperinsulinemia can occur in the absence of insulin resistance and was found to be the strongest predictor of T2D in a large longitudinal study (102, 103). In addition, prolonged insulin administration in both humans and rodents leads to insulin resistance (104–108), suggesting that hyperinsulinemia can induce insulin resistance. This is supported by studies highlighting mechanisms by which hyperinsulinemia can contribute to insulin resistance through downregulating and desensitizing insulin receptors in multiple tissues (49, 109–114) and cell types (115–117). Another possible interpretation is that chronically higher insulin levels lead to a physiological response to downregulate insulin sensitivity to maintain normal glucose utilization and possibly prevent glucose toxicity. Further research into whether chronic insulin administration provokes physiological versus pathophysiological adaptations in insulin sensitivity/resistance is needed. However, in that same theme, studies have suggested hyperinsulinemia can lead to increased adiposity, with prolonged insulin administration associated with weight gain in humans (118), whereas administration of the insulin-lowering drugs diazoxide and octreotide leads to weight loss (119–121). The most direct evidence highlighting the causative role of hyperinsulinemia in insulin resistance and obesity comes from mice genetically modified to be incapable of becoming hyperinsulinemic. These mice do not develop diet-induced obesity (122), have improved insulin sensitivity with age, are protected from obesity-related pathologies and live longer (50). In addition, reducing insulin production in mice with established obesity leads to weight loss without affecting glucose tolerance (101). These findings demonstrate that hyperinsulinemia can play a causal role in the development of insulin resistance and obesity. The lower insulin secretion maintained by regular exercise may therefore be a main factor involved in the preventative effects for metabolic disease. In addition, situations that lead to increased insulin secretion and hyperinsulinemia, such as exercise cessation, may lead to increased insulin resistance and adiposity and increase the risk of developing T2D.

EXERCISE AND β-CELL PROLIFERATION

In addition to altering insulin secretion, exercise has been linked with β-cell proliferation. Studies examining the effect of exercise on β-cell proliferation in rodents and potential mechanisms have been extensively reviewed by Curran et al. (123). Briefly, exercise can increase β-cell proliferation in healthy, prediabetic and diabetic rodent models (123–128). This may only occur in younger animals however, as these studies were carried out in 6–8 wk old rats (124, 127), 7 wk old mice (128), and 5 mo old gerbils (126). Specifically, young β-cells are still capable of substantial adaptive proliferation and apoptosis for physical remodeling (129), meaning they can adapt to their β-cell mass. Conversely, old β-cells are highly refractory to proliferation (130, 131) and have distinct transcriptomic profiles (132), requiring that cell-autonomous adaptations occur through physiological adaptations in older mice rather than hyperplasia. There is also limited research into how cessation of regular exercise impacts β-cell proliferation. Future research into β-cell proliferation levels in older rodents and following exercise cessation may help to understand the effects of exercise and inactivity in different conditions on insulin secretion.

EXERCISE AND β-CELL PROTECTION

Exercise can also have long-term protective effects on human and rodent β-cells (133), including protection from endoplasmic reticulum (ER) stress and other forms of damage via STAT3 activation (134) and/or insulin receptor substrate 2 (IRS2) (98, 127, 135). Exercise appears to be protective for β-cells in models of T1D (136, 137), T2D (138–141), pre-diabetes (128), obesity (142–145), and inflammation (146–148). The prevention of diabetes in ZDF rats by exercise can be phenocopied with AMPK activation (141), suggesting a possible role for this universal cellular energy sensing mechanism. Beyond prevention, exercise has been shown to be an effective treatment for T2D, either alone or in combination with pharmacotherapy (149). In pancreatectomized rats, exercise reduces apoptosis through enhanced IRS2 expression through the activation of cAMP responding element binding protein leading to an increased expression of pancreatic and duodenal homeobox factor-1 (PDX1) (150). Endurance exercise has also been found to reduce the expression of apoptotic proteins and enhance the antioxidant capacity of β-cells in rats (127). The protective effects of exercise on β-cell function further highlight that exercise can lower the risk of developing T2D. Similar to research into β-cell proliferation, no studies have examined the effect that exercise cessation has on β-cell apoptosis. Research into this area could help to uncover how long the protective effects of exercise last, following the cessation of regular exercise, and the contribution apoptosis has in the rapid onset of metabolic dysfunction that can occur following exercise cessation.

MECHANISMS BY WHICH EXERCISE MAY ALTER β-CELL FUNCTION

Alterations to the GSIS Pathway

GSIS requires glucose entry through constitutive glucose transporters, phosphorylation by glucose kinase, metabolism in glycolysis, and oxidative phosphorylation, followed by the closure of K_ATP channels (151). The resulting action potentials cause Ca²⁺ influx via voltage-gated Ca²⁺ channels, which interacts with soluble N-ethylmaleimide-sensitive factor attachment protein receptor (SNARE) proteins resulting in insulin granule exocytosis (151, 152). Exercise-induced alterations in this pathway have been examined to potentially explain the changes in insulin secretion observed following chronic exercise. Some studies have suggested that a decrease in GLUT2 and glucokinase activity may be responsible for the decrease in insulin secretion observed due to exercise training (86, 153). These findings are in contrast to other studies, however, which have reported a decrease in insulin secretion yet no difference in these measures (87) or an increase in glucose oxidation and utilization (83), GLUT2 expression (80), and activity of anaplerotic enzyme levels (79). These latter findings would suggest that adaptations to glucose uptake and utilization are not responsible for decreased insulin secretion following exercise training.

One potential mechanism exercise may alter the GSIS pathway is through a reduction in amplification pathways, which increase the sensitivity of insulin secretion to signals from glucose metabolism and calcium influx. A decrease in amplification pathways have been observed in trained Wistar rats who also had a decrease in GSIS (88). Furthermore, endurance exercise training in rats increased islet levels of phospho-AMPK, CaMKII, and uncoupling protein 2 (UCP2), which were associated with a decrease in GSIS (80). As reviewed by Fu et al. (154), activation of AMPK in β-cells is believed to inhibit GSIS through a number of mechanisms including alteration of glucose metabolism, reduced K_ATP trafficking to the membrane, reduced insulin gene transcription, and reduced insulin granule docking (154). Activation of β-cell AMPK may, therefore, be a key signaling molecule in the lower insulin secretion observed with regular exercise (Fig. 2). In addition, an increase in the expression of UCP2 may reduce insulin secretion by uncoupling oxygen consumption to ATP synthesis. AMPK activation, increased UCP2, and a decrease in amplification pathways may explain why a decrease in GSIS was observed in trained rodents even in the context of increased glucose uptake and utilization (83). Further research into these pathways following exercise training and exercise cessation is needed to add to limited data and help understand the underlying mechanisms of how insulin secretion is altered by exercise.

Figure 2.

Potential mechanisms exercise and cessation of exercise impact β-cell function. Exercise and the cessation of regular exercise may alter a number of molecular mechanisms that can adapt β-cell function, survival, proliferation, and redifferentiation. These mechanisms include exerkines released from tissues such as skeletal muscle and liver, glucocorticoids released from the adrenal gland, incretin release from the intestine and pancreatic α-cells, stimulation from the branches of the autonomic nervous system, AMPK activity, amplification pathways, insulin clearance and contraction-mediated glucose uptake altering insulin levels, and β-cell insulin sensitivity. Ach, acetylcholine; AMPK, AMP-activated protein kinase; BAIBA, β-aminoisobutyric acid; CX3CL1, fractalkine; GIP, glucose-dependent insulinotropic peptide; GLP-1, glucagon-like peptide-1; GSIS, glucose-stimulated insulin secretion; IDE, insulin-degrading enzyme; IL6, interleukin 6; InsR, insulin receptor; NE, norepinephrine; PSNS, parasympathetic nervous system; SNS, sympathetic nervous system. Created with Biorender.com.

EXERKINE SIGNALING

Exerkines are signaling molecules released into circulation during exercise that may act locally or on other tissues (155). A number of studies and reviews have highlighted a potential role for exerkines to induce some of the beneficial β-cell adaptations to exercise (155–158). In vitro studies indicate that the effects of exercise on β-cells may be transmitted to pancreatic islets via soluble factors. In mouse islets and rat and human β-cells, myotube-conditioned media increased GSIS at a high glucose concentration of 16.7 mM (159, 160). In addition, conditioned media from contracting C2C12 myotubes increased mitochondrial activity and potentiated insulin secretion from the INS-1 β-cell line at both 5 mM and 20 mM glucose (158). These findings contrast with most studies in vivo and in isolated islets from trained rodents, which indicate exercise training decreases insulin secretion. These differences are likely linked to exerkine studies mimicking “during exercise” responses whereas islet studies from chronically exercised rodents are studying programmed adaptations in a nonexercise condition. A subsequent study, however, did find that conditioning mouse islets with serum from trained mice decreased insulin secretion at a high glucose concentration (161) and sera taken following exercise training in humans can reduce insulin secretion in INS-1 cells (157). These contrasting results may be due to the numerous other factors in vivo that may affect myokine levels and insulin secretion and highlight that myokines may not be the main driver behind a decrease in insulin secretion observed following exercise training. Exerkines from tissues other than skeletal muscle may also play a role in inducing β-cell adaptations to exercise.

Muscle-derived exerkines may, however, play an important role in other β-cell adaptations to exercise training such as increased proliferation and β-cell protection. Conditioned media from resistance trained mice was found to prevent markers of ER stress in the INS-1E β-cell line (93). Similarly, human islets and mouse and rat β-cells have reduced apoptosis when exposed to inflammatory cytokines or ER stress inducers if cultured in media with sera from endurance-trained humans or mice (134). In addition, myotube conditioned media has been found to increase proliferation of rat β-cells in vitro (160). A potential indirect mechanism in which exercise training may be beneficial to β-cell health is through the prevention of insulin resistance in muscle. Conditioned media from myotubes made insulin resistant through TNF-α treatment significantly increased apoptosis in human and rat β-cells and decreased proliferation in rat β-cells (160). The circulating factors that link exercise to β-cell function and survival remain to be conclusively identified, but several candidates have been proposed (157, 162).

IL6

IL-6 has been one of the most widely studied exercise-induced myokines and has been proposed to potentiate insulin secretion indirectly via stimulating glucagon-like peptide-1 (GLP-1) secretion from islet α cells (163), although the role of GLP-1 remains somewhat controversial (164). IL-6 was identified as a key factor in the protective effects of C2C12-conditioned media in an in vitro model of type 1 diabetes-associated β-cell stress (161) and in the protective effects of media conditioned with sera from exercised mice or humans on human islets and rat and mouse β-cells (134). The protective effects of IL-6 has been reported to be context-dependent, with IL-6 exposure decreasing β-cell apoptosis under normal conditions but increasing apoptosis in proinflammatory conditions (157) and some studies have failed to find significant effects (165).

CX3CL1 (Fractalkine)

The chemokine CX3CL1, also known as fractalkine, is a myokine with elevated plasma levels following exercise (166–168) and has been suggested to play a role in β-cell function (169). The study of CX3CR1 KO mice suggests that CX3CL1 potentiates GSIS through MEK-dependent Ca²⁺ mobilization (170). This is supported by findings that CX3CL1 enhances GSIS in isolated mouse islets and in HFD mice following chronic administration of CX3CL1 (171). It has also been demonstrated that CX3CL1 can protect against apoptosis in TNFα-treated human islets, palmitate-treated mouse islets, and HFD mice (171, 172). This may be through inhibiting the expression of inducible cAMP early repressor-1 (ICER-1), a transcriptional repressor that is expressed due to an HFD and impairs β-cell function (170, 171). It has also been suggested however, that elevated CX3CL1 in humans can increase the risk of cardiovascular disease and diabetes (173, 174). Further investigation into the role of CX3CL1 as an exerkine in inducing β-cell adaptations to exercise and exercise cessation could help to clarify the importance of this molecule.

Follistatin

Circulating levels of follistatin significantly increase both during and following exercise and are putatively released by the liver (175–177). This may have an effect on β-cell function as elevated follistatin has been shown to potentiate insulin secretion by re-differentiating mouse and human β-cells (178). Follistatin may possibly be an exercise-regulated mediator of β-cell function (176) but more research is clearly needed.

Irisin

Irisin is another putative exercise-induced myokine that may have a role in the β-cell adaptations to exercise (179). In INS1-E cells and isolated mouse islets incubated in glucolipotoxic conditions, irisin prevents apoptosis and the accumulation of lipids in β-cells (180–183). This is associated with an increase in AMPK (180) and prosurvival regulation of the expression of apoptotic proteins such as B cell lymphoma-2 (BCL-2), BCL-2 associated X protein (BAX), and caspase-3 (181–183). Irisin is potentially important in the context of protecting β-cell in metabolic disease. Studies have also found irisin to increase the proliferation of INS-1 cells and human pancreatic 1.1B4 cells in vitro (181, 182), with increased β-cell proliferation also observed in mice administered irisin to raise levels 1.5 fold (181). Exposure to irisin also increases GSIS both in vitro, from human and murine islets and INS-1E cells, and in vivo, as observed in C57BL/6 mice (181, 183). Although irisin can be increased due to exercise, elevated levels have also been observed in the context of obesity (181, 184). Exercise may therefore provide a means to receive the potential benefits to β-cell function outside of the context of metabolic disease.

β-Aminoisobutyric Acid

β-Aminoisobutyric acid (BAIBA) is a metabolite released from skeletal muscle, with plasma levels increasing during exercise (185, 186) and elevated in aerobically trained humans (187). Insulin secretion is decreased in INS-1832/3 cells treated with BAIBA at 5 mM glucose suggesting BAIBA may directly interact with β-cells to alter insulin secretion (185). This may not be physiologically relevant however, as the increase in circulating BAIBA due to exercise is ∼250 nM (186) whereas decreased insulin secretion in INS-1832/3 cells is only observed at concentrations of 10 µM and higher (185). Further studies in isolated pancreatic islets may also provide additional evidence on the role BAIBA has in modulating β-cell function.

Fetuin-A

Fetuin-A is a glycoprotein released from the liver, while not an exerkine, its circulating levels decrease with exercise (188–191). Raised fetuin-A can significantly decrease GSIS at high glucose concentrations (192, 193), whereas other studies have found it to increase basal insulin secretion in human pancreatic islets (194). In addition, fetuin-A elevates apoptosis and lipotoxicity of β-cell lines in vitro (193, 195) and disrupts β-cell maturation (194). Exercise-induced decreases in circulating fetuin-A may therefore be involved in the protective effects seen for β-cells.

Glucocorticoids

Intermittent exposure to high levels of glucocorticoids from regular exercise is also believed to play a role in linking exercise and β-cell function (196). Interestingly, 13 wk of intermittent stress induced by restraint (1 h/day, 5 times per wk) in ZDF rats preserved β-cell mass (197). Similar adaptations are observed in ZDF rats following exercise training (124), suggesting that the glucocorticoid exposure due to regular exercise may be involved in maintaining β-cell function in T2D. Some of the beneficial effects of glucocorticoids on β-cells are thought to occur due to increased PDX-1 activity, promoting β-cell survival and inhibiting forkhead box protein O1 (FOXO-1) induced apoptosis (196).

Further Exerkine Research

The effects of different exerkines on GSIS are mixed and ambiguous. Although many exerkines have been reported to enhance GSIS, this is often not observed in studies following exercise training in humans and rodents (Table 1). The main effect of many exerkines may therefore be on β-cell protection and proliferation, with a number of exerkines mentioned earlier having the ability to reduce β-cell apoptosis under high glucolipotoxic conditions and stimulate β-cell proliferation. However, the cause-and-effect relationships between putative exerkines and exercise-induced insulin modulation are unclear. In addition, many of the exerkine β-cell function studies have only seen effects at supraphysiological concentrations. Further research into the role of exerkines in vivo following exercise training and cessation of exercise could therefore provide important insight into the mechanisms by which these physiological conditions effect β-cell function.

AUTONOMIC NERVOUS SYSTEM TONE

Exercise can modulate insulin secretion by acutely increasing sympathetic tone (89), a likely cause for the rapid lowering of insulin during a bout of exercise (198–200). As reviewed in detail by Ahrén et al. (201), neurotransmitters released from sympathetic terminals that innervate the pancreas can inhibit GSIS (201). These include norepinephrine (NE), which inhibits GSIS by binding to α₂-adrenoreceptors leading to hyperpolarization of the β-cell plasma membrane, preventing the closing of K_ATP channels and Ca²⁺ uptake (202). Inhibiting the α₂-adrenoreceptor during exercise has been shown to prevent the decrease of insulin secretion (200). In addition, galanin and neuropeptide Y (NPY) can inhibit GSIS through hyperpolarization of the plasma β-cell plasma membrane and decreasing the production of cAMP (201, 203).

Less is known on whether regular exercise, or the cessation of regular exercise, leads to adaptations in resting pancreatic autonomic innervation that contribute to altered insulin secretion. A study by Gomes et al. (89) found that reduced insulin secretion in exercise-trained rats was associated with an increase in sympathetic nervous system activity. Most studies examining the link between autonomic nervous system activity and exercise training in humans however conclude that exercise training decreases sympathetic tone and increases parasympathetic tone at rest (204, 205). This highlights that the role of the autonomic nervous system in the β-cell adaptations to exercise may be complex or not play a significant role in insulin secretion at rest. One hypothesis raised by Dela (206) is that regular exposure to NE may contribute to exercise-induced β-cell adaptations that lower insulin secretion. Following 10 days of a 45 min daily infusion of NE, to mimic the increase seen during exercise, insulin secretion during a hyperglycemic clamp was significantly reduced in men (206). This suggests that β-cells adaptations due to exercise may occur due to intermittent NE exposure however, the mechanisms behind this remain unclear.

Exercise exposure may induce regional differences in sympathetic neural tone. This has been observed in the different renal and cardiac sympathetic activity observed due to chronic exercise (207, 208). Although sympathetic innervation of the cardiovascular system may decrease due to exercise training (209, 210), a different response may be observed in sympathetic neural activity in the pancreas. In rats fed a chow diet, those with access to a running wheel for 4 to 16 wk show lower splanchnic nerve activity compared with sedentary rats (211–213). Rats with obesity induced by postnatal early overfeeding, however, have higher splanchnic nerve activity following exercise training compared with sedentary rats (214). These differences suggest that exercise may modulate the activity of the splanchnic nerve differently depending on metabolic health. It is also possible that methodological differences in rodent studies, such as the duration postexercise before measurement of the splanchnic nerve activity, are responsible for these observations.

Another potential mechanism by which alterations in autonomic innervation may alter insulin secretion is through a decreased responsiveness to autonomic neurotransmitters. Indeed, the effect of NE on inhibiting GSIS appears to diminish with training, with the inhibitory effect of NE not observed in trained mice due to reduced inhibition of cAMP levels (84). In addition, the stimulatory effect of acetylcholine on insulin secretion is reduced in islets isolated from exercised rodents (88). This decrease in responsiveness to neurotransmitters may be due to a decreased expression of β-cell neurotransmitter receptors. At rest, where there may be a greater influence of the parasympathetic nervous system on the pancreas, a decrease in responsiveness to acetylcholine could therefore be involved in the decreased GSIS observed following exercise training.

Autonomic nervous system tone may also play a role in β-cell proliferation. As reviewed by Moullé (215), parasympathetic innervation of islets can increase β-cell proliferation while sympathetic innervation can decrease β-cell proliferation. An increase in resting parasympathetic tone due to chronic exercise training may therefore have beneficial effects by contributing to increased β-cell proliferation. This may be particularly important during adolescence when the pancreas has a higher plasticity for β-cell proliferation and may highlight the importance for regular exercise and/or PA in adolescence for long-term programming of β-cell mass. Further research on this topic could help to address this hypothesis.

The potential role of neural innervation of the pancreas in altering insulin secretion due to exercise is questionable, however, as some studies show that insulin secretion during exercise is not altered due to denervation of the pancreas (216). In addition, human islets have little neural innervation compared with rodent islets, which may explain contrasting results (215, 217). Furthermore, the fact that changes in GSIS are observed in isolated islets suggests other β-cell adaptations are involved in these changes. In conclusion, there are many questions to be answered on how the autonomic nervous system modulates pancreatic activity (218).

INSULIN CLEARANCE

Exercise has been found to increase insulin clearance in mice (219, 220) and humans (221–228). Elevated hepatic levels of insulin-degrading enzyme (IDE) have also been found in exercised rats (229). In addition, exercise cessation for 6 to 7 days can lead to a decrease in insulin clearance (230, 231). Exercise-induced increases in IL-6 may be involved, as IL-6 has been shown to increase IDE levels in HEP2G and C2C12 cells and IL-6 knockout mice have reduced IDE levels and activity (232). Changes in insulin clearance may be context dependent, with other studies finding exercise to have no effect on insulin clearance (233).

Alterations in insulin clearance due to exercise may indirectly influence β-cell function through preventing hyperinsulinemia. Hyperinsulinemia may feedback onto the β-cell, downregulating the insulin receptor and reducing the negative feedback of insulin-on-insulin secretion. This effect of insulin has been previously reported in humans (234) and recent data supports this by showing that specific insulin receptor knockout in β-cells leads to increased insulin secretion (235). Studies that experimentally manipulate insulin clearance, through modulation of IDE, are needed to understand the role of insulin clearance on fasting and postprandial insulin concentrations with or without regular exercise

INCRETINS AND EXERCISE

Incretins are hormones released following the ingestion of nutrients and include GLP-1 and glucose-dependent insulinotropic peptide (GIP). Incretins have an important role in glucose homeostasis by amplifying insulin secretion. A final mechanism by which exercise may effect β-cell function is through modulating the expression of the incretins GLP-1 and GIP. A recent meta-analysis of 16 studies identified that both short- and long-term training interventions increase GLP-1 levels in participants who are healthy, obese, or have T2D (236). This finding is interesting given that GLP-1 is known to potentiate GSIS (237), however, exercise training is associated with a decrease in GSIS in most studies. Interestingly, while insulin secretion from isolated islets from sedentary rats fed a chow or HFD diet exposed to low glucose concentrations is increased with exposure to GLP-1, by 75% and 100% respectively, the exposure of isolated islets from swim-trained rats to GLP-1 leads to a 21% decrease in insulin secretion in chow-fed rats and only a 17% increase in HFD rats (78). A higher increase in insulin secretion at high glucose concentrations was, however, observed in swim-trained rats compared with sedentary rats. The potentiating effect of GLP-1 at low glucose concentrations may, therefore, be blunted following exercise training and may be associated with the lower insulin secretion observed in trained humans and rodents. Indeed, it has been hypothesized that a decrease in the responsiveness of β-cells to GLP-1 is responsible for the decrease in insulin secretion observed in trained subjects that do not see a change in postprandial GLP-1 secretion following an exercise training program (238). This may not occur at high glucose concentrations, suggesting other factors are likely responsible for the decrease in GSIS in response to high glucose. Like exercise, GLP-1 can increase β-cell proliferation in rodents (239). The increase in GLP-1 following exercise could therefore play a role in the beneficial effects of exercise on β-cell proliferation.

GIP has also been suggested to be involved in the changes in insulin secretion observed following exercise training. Following a diet and exercise intervention, GIP and insulin secretion increased in patients with T2D, whereas a nonsignificant decrease (P = 0.07) in GIP and a significant decrease in insulin secretion was observed in obese subjects with normal glucose tolerance (240). Context-dependent changes in GIP release following exercise training provide a potential mechanism behind the context-dependent changes in insulin secretion observed in humans and rodents. The role of exercise in inducing changes in GIP levels is questionable, as a number of studies have found neither acute, nor chronic exercise, has a significant impact on GIP secretion (67, 241–243). It is also important to note that as decreases in insulin secretion following exercise training are detected using the hyperglycemic clamp technique, which bypasses the splanchnic system, potential exercise training-induced alterations in enteroendocrine incretin secretion may not play a significant role in altering GSIS. Pancreatic α-cell incretin release may, therefore, be responsible if alterations in incretin release or stimulation play a main role in adaptations to insulin secretion. Incretin levels following exercise cessation and their impact on insulin secretion also remain to be investigated.

CONCLUSIONS

Collectively, the observations discussed in this review highlight that exercise leads to pancreatic β-cells adaptations including modulation of glucose-stimulated insulin secretion, increased proliferation, and protection from stressful stimuli known to negatively influence β-cell function. In most studies of humans and rodents with normal β-cell function, exercise training decreases insulin secretion, which may have an important role in preventing hyperinsulinemia and the development of metabolic dysfunction. In circumstances of reduced β-cell function, exercise may induce a beneficial increase in insulin secretion, however, a retained level of underlying function may be required. This has ramifications for when exercise interventions are used to preserve β-cell function in individuals transitioning from prediabetes to type 2 diabetes. Cessation of regular exercise also leads to a rapid increase in insulin secretion, suggesting the beneficial effects of exercise on insulin secretion may be relatively short lived.

The exact mechanisms that drive these β-cells adaptations to exercise remain to be confirmed and it is likely that numerous factors are in play. Potential mechanisms that have been identified include activation of β-cell AMPK, signaling from exercise-induced exerkines, altered stimulation/response to the autonomic nervous system, intermittent exposure to NE and glucocorticoids, altered insulin clearance, and an alteration in the incretin response (Table 2). Despite the critical role that insulin has on metabolic health, limited research has been conducted into alterations to these pathways and β-cell function following exercise cessation. In addition, most research into these mechanisms has used moderate to vigorous aerobic exercise as a stimulus and little is known on the effects of other forms of exercise, such as resistance and anaerobic training, on pathways potentially responsible for adapting β-cell function. The underlying molecular pathways for a number of these mechanisms, and how they may lead to context-dependent β-cell adaptations to exercise and cessation of regular exercise, provides a promising area for future study. Increased understanding of how exercise is beneficial to the pancreatic β-cell may help in the development of therapeutics for those that are limited in participating in exercise or moderate to vigorous PA.

Table 2.

Hypotheses for lower insulin secretion with regular exercise

Activation of AMPK in the β-cell

Upregulated UCP2 in the β-cell

Reduction in GSIS amplification pathways

Blunted response to autonomic nervous system regulation, possibly leading to lower insulin secretion by reducing the effect of acetylcholine

An adaptation received from intermittent exposure to NE and glucocorticoids

Improved insulin clearance, preventing hyperinsulinemia and desensitization of β-cell insulin resistance

Blunted GLP-1 response at low glucose

Decreased GIP levels, while increased GIP levels in T2D improve insulin secretion

Exposure to increase BAIBA levels

AMPK, AMP-activated protein kinase; BAIBA, β-aminoisobutyric acid; GIP, glucose-dependent insulinotropic peptide; GLP-1, glucagon-like peptide-1; GSIS, glucose-stimulated insulin secretion; T2D, type 2 diabetes; UCP2, uncoupling protein 2.

Challenges for the field going forward will be to resolve the short-term and long-term effects of exercise cessation, and to define the physiological and molecular mechanisms involved using integrative in vivo models where possible. Collectively, we will need to do a better job of using more physiologically relevant housing, studying both sexes, and a range of age and disease condition (healthy vs. insulin resistant vs. T2D, etc.) groups to better model more relevant conditions.

GRANTS

L.G.H. and J.D.J. were supported by the Canadian Institutes of Health Research (CIHR) under Grant No. PBSM GR015635. J.P.T. was supported by the NIH under Grant No. R01 DK121497.

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the authors.

John Thyfault is an editor of Journal of Applied Physiology and was not involved and did not have access to information regarding the peer-review process or final disposition of this article. An alternate editor oversaw the peer-review and decision-making process for this article.

AUTHOR CONTRIBUTIONS

L.G.H. and J.P.T. prepared figures; L.G.H., J.P.T., and J.D.J. drafted manuscript; L.G.H., J.P.T., and J.D.J. edited and revised manuscript; L.G.H., J.P.T., and J.D.J. approved final version of manuscript.