Brain insulin resistance and cognitive function: influence of exercise

Abstract

Exercise has systemic health benefits in people, in part, through improving whole body insulin sensitivity. The brain is an insulin-sensitive organ that is often underdiscussed relative to skeletal muscle, liver, and adipose tissue. Although brain insulin action may have only subtle impacts on peripheral regulation of systemic glucose homeostasis, it is important for weight regulation as well as mental health. In fact, brain insulin signaling is also involved in processes that support healthy cognition. Furthermore, brain insulin resistance has been associated with age-related declines in memory and executive function as well as Alzheimer’s disease pathology. Herein, we provide an overview of brain insulin sensitivity in relation to cognitive function from animal and human studies, with particular emphasis placed on the impact exercise may have on brain insulin sensitivity. Mechanisms discussed include mitochondrial function, brain growth factors, and neurogenesis, which collectively help combat obesity-related metabolic disease and Alzheimer’s dementia.

INTRODUCTION

Approximately 42% of American adults were defined as obese according to their body mass index (BMI) in 2017–2018, and it is projected that this will increase to 48.9% by 2030 (1, 2). The rise in obesity is alarming because it coincides with a greater risk for type 2 diabetes (T2D), in part, through insulin resistance. Traditionally, insulin resistance is defined as an inability of insulin to exert regulation of blood glucose via “below the neck” mechanisms. For instance, decreased skeletal muscle glucose uptake and/or elevated hepatic glucose production as well as lipolysis are considered to promote stress on the pancreas to overproduce insulin in an effort to maintain circulating glucose within normal ranges. If this underlying insulin resistance persists, the pancreas becomes “exhausted” to the point of generating little or no insulin, thus creating the hyperglycemic state seen in T2D (3). More recently, elevated glucose levels and insulin resistance have been linked to neuropsychological and neurologic diseases in humans, including major depression, cognitive decline, and Alzheimer’s disease (AD) (4–6). In fact, AD has even been coined “type 3 diabetes” (7). These observations have led the field to examine the role of insulin and glucose homeostasis beyond T2D to elucidate dynamic bidirectional pathways between the brain and periphery that have been implicated in declines in cardiovascular and cognitive health.

Given that ∼5 million Americans are living with AD and almost 35 million with T2D, understanding the interaction between these two diseases has clinical implications for treatment and prevention (8, 9). Indeed, although the brain only comprises ∼2% of total adult body weight, it uses upward of 20% of blood glucose as an energy source (10). Much of this glucose utilization by the brain is noninsulin dependent, leading many to believe the brain is not reliant on insulin action for general health. However, more recent work in rodents and humans highlights that the brain is an insulin-sensitive organ as evident by the presence of insulin receptors/signaling cascades (11, 12). This action of insulin in the brain is also important for roles in cognition and feeding behavior (13, 14). Interestingly, exercise has gained attention as an effective behavioral prevention and/or treatment not only for T2D but also for minimizing cognitive decline in neurotypical aging and among patients with AD (15–17). Herein, we discuss the relationship between peripheral and brain insulin resistance and recent evidence that examines the effect of insulin on mitochondrial metabolism, brain structure and function, and cognition. In this review, we focus on how insulin and other growth factor-related hormones [i.e., insulin-like growth factor (IGF-1) and brain-derived neurotrophic factor (BDNF)] influence the brain to modulate both glucose homeostasis as well as cognition. We also review the literature on the effect of exercise on brain structure and function as it relates to age-related cognitive decline and AD from rodent and human studies, with particular emphasis on how brain insulin sensitivity may be targeted as a novel mechanism for promoting healthy cognitive function through exercise therapy (Fig. 1). Finally, we discuss the clinical implications of using medications including intranasal insulin and insulin sensitizers as an alternative or adjunct to exercise for overall well-being.

Figure 1.

A novel mechanism for promoting healthy cognitive function with exercise. Created with BioRender.com with permission. AD, Alzheimer’s disease; ROS, reactive oxygen species.

“TRADITIONAL” INSULIN RESISTANCE

Blood glucose regulation is traditionally considered to be controlled through cross talk interactions of the pancreas, liver, skeletal muscle, adipose tissue, and vasculature (18). Under fasting conditions, the α cells of the pancreas secrete glucagon, a glucoregulatory peptide hormone, responsible for stimulating endogenous glucose production to maintain normoglycemia. The liver, with secondary influences from the kidney and to a lesser extent the small intestine, is primarily responsible (e.g., ∼80%–85%) for maintaining fasting plasma glucose (18). Thus, fasting plasma glucose is clinically used to crudely estimate hepatic glucose metabolism. Under fed conditions though, insulin levels typically rise after carbohydrate absorption, thereby leading to reductions in hepatic glucose production and lipolysis while increasing skeletal muscle blood flow for insulin-mediated glucose uptake (18). Subsequently, hyperglycemia is believed to develop when pancreatic insulin secretion fails to overcome insulin resistance in either the liver, skeletal muscle, or adipose tissue. How exactly this peripheral insulin resistance develops is an area of intense investigation, but it is often attributable to excess nutrients promoting oxidative stress/inflammation that, in turn, impair the action of insulin to promote glucose uptake (19). Research from the past several years has also shown the key roles of insulin and insulin signaling in the brain (20) (see INSULIN ACTION ON BRAIN METABOLISM). The notion that the pancreas is also key for brain insulin action is worth noting as the debate has occurred regarding the ability of neurons and glial cells to produce insulin in the nervous system, although clear evidence is currently lacking. Thus, insulin transport from the blood into the brain becomes relevant and passes through the blood-brain barrier (BBB) and/or blood-cerebrospinal fluid barrier (21). However, studies have questioned diffusion through the use of radiolabeled insulin and identified that insulin localizes to brain endothelial cells in various regions (e.g., hippocampus, hypothalamus, and frontal cortex) (22, 23). In fact, insulin receptors on endothelial cells seem to have a key role in the regulation of insulin transport throughout the body. This is believed to occur in a variety of tissues (e.g., skeletal muscle and brown fat) with continuous or tight endothelial barriers when compared with other tissues with discontinuous endothelial barriers (e.g., liver) (21). The brain is unique though in that the BBB is a combination of endothelial cells, pericytes, and glial cells that collectively form tight junctions between endothelial cells and limit insulin delivery. To this extent, it is of interest that in a knockout model of endothelial insulin receptors, insulin stimulation had delayed responses for insulin signaling in olfactory bulb compared with hypothalamic, hippocampal, or prefrontal cortex neurons (21). This is clinically and scientifically germane for human health in understanding the relation of pancreatic insulin section and insulin action on the brain for regulating food intake, glucose metabolism, and cognitive benefit.

INSULIN ACTION ON BRAIN METABOLISM

Although the central nervous system regulates energy homeostasis (24), less attention has been directed toward the action of insulin in regulating brain function. Notably, the insulin receptor is expressed widely in the brain with the highest levels seen in the cerebellum, cortex, and hypothalamus (25). Upon entry into the brain, insulin is known to activate canonical pathways including IRS-1 and IRS-2 thereby leading to the activation of PI3K, AKT, and so forth. In particular, activation of AKT is known to stimulate the mammalian target of rapamycin c (mTOR) that impacts synaptic plasticity, neurotransmitter trafficking, and neuronal survival (26–28). A key difference in insulin action in the brain relates to glucose uptake. First, peripheral insulin receptors are mainly in the α isoform compared with predominantly β isoforms in the brain, which is important considering the higher insulin binding affinity in the former isoform (21). Second, neurons, glial cells, and brain endothelial cells rely on GLUT1 and GLUT3 transporters compared with mostly GLUT4 transporters in skeletal muscle under insulin-stimulated states (29). Importantly, brain glucose transporters are mainly insulin-independent compared with GLUT4. Although this makes good evolutionary sense for having a low affinity for glucose to be utilized by the central nervous system in times of need, GLUT4 transporters are located heavily in the cerebellum, hypothalamus, and hippocampus (30). Thus, the combined work indicates that most of the glucose uptake in the brain per se is not regulated by insulin signaling. This points toward the notion that insulin resistance in the brain is related more to impaired insulin signaling pathways per se. Indeed, peripheral insulin resistance often leads to “below the neck” hyperinsulinemia that can downregulate central nervous system insulin signaling and contribute to the formation of tau phosphorylation, amyloid-β toxicity, and oxidative stress/inflammation that jointly, or independently, raise cognitive decline risk (31). In addition, deficient brain insulin signaling may interfere with lipid transportation (notably cholesterol) as well as protein expression (e.g., GABA, NMDA, etc.) linked to depression, cognitive decline, and dementia (31). Together, it seems insulin plays a unique role in modifying brain metabolism pathways. In fact, there is good evidence from rodent studies that insulin acts on PI3K to stimulate mTOR, which is a known regulator of PGC1-α and nuclear respiratory factors 1 and 2 for mitochondrial biogenesis and metabolism as well (21). mTOR is known to stimulate mitochondrial proteins involved in the TCA cycle, fatty acid β-oxidation, as well as the electron transport chain (32). This is important as mitochondria are a key producer of cellular ATP, oxidative stress, and energy homeostasis. Given that obesity and T2D promote mitochondrial dysfunction, it is not surprising that insulin resistance is a related characteristic.

INSULIN REGULATES BRAIN SYSTEMIC PHYSIOLOGY

One of the established effects of insulin on the brain relates to the regulation of feeding. Insulin is known to reduce food intake in part through the binding on insulin receptors of the pro-opiomelanocortin (POMc) and agouti-related protein (AgRP) neurons in the arcuate nucleus of the hypothalamus to induce satiety as well as blunt orexigenic behaviors, respectively (33, 34). Indeed, using intranasal insulin as a treatment has been shown to induce fat loss in men, but not in women, over an 8-wk timeframe (35). Although this work suggests potential sex-based differences in insulin regulation of food intake, the work highlights that brain insulin signaling influences peripheral homeostasis. In addition to this classic action of insulin, it is also recognized to stimulate sympathetic outflow to regulate body temperature as well as increase postprandial thermogenesis (12). Insulin also affects peripheral tissues in maintaining substrate concentration levels. Indeed, insulin acts on the neurons that express AgRP neurons from rats to regulate hepatic glucose production (36, 37), while at the same time influencing POMc neurons to inhibit adipose tissue lipolysis in mice (38). In fact, decreases in hypothalamic lipid oxidation in the fasted state inhibit glucose production in the liver by activating ATP-sensitive potassium (K_ATP) channels in the ventromedial hypothalamus (VMH). From a neural perspective, this partly explains why hepatic glucose production decreases under fed conditions since the preferred metabolic fuel source switches from fat to carbohydrates in the brain. It is important to note that the VMH, along with the arcuate hypothalamic nucleus (ARC), also influences glucose uptake in skeletal muscle and adipose tissue. Toda et al. (39) demonstrated that leptin injection into the VMH increased glucose uptake in skeletal muscle, brown adipose tissue, and the heart in lean mice. However, the effects of insulin on hepatic glucose production and insulin sensitivity have been controversial with mixed effects reported. To further elucidate the effect of insulin on the central nervous system in women, acute intranasal insulin administration (160 IU) reduced appetite in the postprandial state (40) despite no significant differences in blood glucose concentrations. This is consistent with previous studies showing no significant decrease in hepatic glucose production following acute intranasal insulin administration in humans (41–43). However, Kishore et al. (44) showed that K_ATP activation via acute oral administration of diazoxoide (a known hypothalamic K_ATP activator) using the pancreatic euglycemic-hyperinsulinemic method significantly decreased hepatic glucose production by ∼30% in humans. Dash et al. (45) reported similar findings in healthy men where acute intranasal insulin administration (40 IU) during a pancreatic clamp significantly decreased hepatic glucose production by ∼35% compared with placebo. However, in a follow-up study with overweight/obese insulin-resistant men receiving the same intranasal insulin dose, others report no differences in hepatic glucose production during a pancreatic clamp between the experimental insulin and control group (43). The Memory Advancement with Intranasal Insulin in Type 2 Diabetes (MemAID Trial) demonstrated that intranasal insulin treatment for 24 wk increased normal and dual task walking, raised cerebral blood flow, and lowered insulin resistance as estimated by Homeostatic Model Assessment of Insulin Resistance (HOMA-IR) in people with T2D (46). Taken together, and despite controversial studies of the brain regulating hepatic glucose production, the combined literature highlights complex interactions between insulin, the brain, and peripheral tissues.

INSULIN AND COGNITIVE FUNCTION

Cerebral glucose uptake is an important mechanism supporting cognition (20). In turn, it is not surprising that raising systemic insulin levels would favor brain glucose uptake and support memory, executive function, and cognition. Conversely, reduced glucose uptake via insulin resistance has been suggested to be an underlying contributor to accelerated brain aging (47). This notion was supported in a recent study where acute intravenous insulin administration (80 mU/m²/min) during a euglycemic-hyperinsulinemic clamp improved working memory and cognition in healthy older adults (48) as evidenced by increased blood oxygen level-dependent (BOLD) signaling via functional MRI (fMRI). Interestingly, it was also reported that diminished task-related activation was associated with decreased insulin sensitivity. In addition, Craft et al. (49) reported that regular intranasal administration of insulin (40 IU/day) for 16 wk improved memory, brain volume through structural MRI, and AD biomarkers in cerebrospinal fluid after just 2 and 4 mo in adults with mild cognitive impairment (MCI). However, not all areas of the brain may respond to insulin. For instance, insulin does not appear to increase BOLD signaling in primary visual cortex (50, 51). This follows earlier work conducted by Banks and Kastin (52) in mice who showed endogenous insulin crosses the BBB and several forebrain structures (e.g., hypothalamus, hippocampus, parietal cortex, and frontal cortex) but does not cross midbrain structures or the occipital cortex (location of primary visual cortex). As a result, the effect of insulin on the brain seems to be localized to regions responsible for higher-level cognitive processes, including executive functions (e.g., frontotemporal regions).

In addition to prefrontal-dependent executive processes, insulin is also a key component of hippocampal-dependent memory processes. For instance, acute (i.e., 1 day) and/or habitual (i.e., 21 days) intranasal insulin has been reported to improve memory for visuospatial and odor-cued spatial memory (53, 54) as well as verbal fluency tasks in adults with obesity and T2D (53, 55). These findings parallel work in young adults, whereby intranasal insulin increased working verbal memory performance 75 min after administration (56). In older adults with T2D, acute intranasal insulin has been reported to modify the functional connectivity among brain regions regulating memory and sensory and affective processing to higher-level cognition (57, 58). However, not all studies support cognitive benefits from habitual intranasal insulin in people with MCI (59), and it is possible that cognitive task specificity as well as the dose and delivery method of insulin could contribute to the observed differences. In either case, brain insulin action is emerging as a reasonable target to improve neurocognitive outcomes and overall quality of life.

BRAIN INSULIN RESISTANCE

The etiology of brain insulin resistance awaits to be fully elucidated, although several mechanisms have been proposed (Fig. 2A). Like obesity-induced insulin resistance in tissues such as skeletal muscle and the vasculature, it is not surprising that a high fat/sugar diet, lack of physical activity, and genetic predisposition have been suggested as modifiable and nonmodifiable risk factors for brain insulin resistance. Indeed, many of these risk factors are related to developing cognitive decline and AD (28). A unifying conceptual theme of these risk factors may relate to and include inflammation and oxidative stress. The elevated nutrient availability (i.e., fatty acids or glucose) may overload mitochondrial flux, such that there is an increase in reactive oxygen species (ROS). Although ROS itself is a key player in cell signaling, energy sensing, and antioxidant processes, the genesis of excessive ROS can be problematic for brain health (60–62). The brain, compared with skeletal muscle, has relatively low antioxidant activity levels (63–65). It is therefore not surprising that brain insulin resistance is linked to oxidative stress as well as elevated levels of protein, lipid, and DNA oxidation byproducts (66), suggesting heightened sensitivity of the brain to oxidative stress than peripheral tissues. In turn, these mechanisms may help to explain the associated declines in insulin receptor expression in brains of patients with AD compared with age-matched controls (67), as well as lower levels of insulin signaling cascade proteins (68). Although the pathogenesis of AD is complex and awaits to be further elucidated, extracellular plaques composed of β-amyloid and accretion of phosphorylated τ proteins into intracellular tangles is a core feature of AD. In turn, it is of interest that insulin receptor knockout animals have an accumulation of τ protein (69). Furthermore, treatment of AD mouse models through insulin pathways increases the clearance of β-amyloid (70). Interestingly, individuals with T2D have mild to moderate reductions in cognition and have an approximate 50% elevated risk of developing dementia compared with nondiabetic controls (71). This evidence insinuates that brain insulin resistance may be an important mechanism underlying cognitive decline in obesity and T2D. Indeed, Baker et al. (72) compared insulin resistance (HOMA-IR), working memory (memory encoding task), and brain activation ([F-18]FDG PET) between nondiabetic adults and adults with prediabetes/early T2D. Not only was insulin resistance associated with reduced glucose uptake in the frontal lobe, like patients with AD, but subjects with prediabetes/early T2D recalled fewer items during a memory recall task and showed diffuse neural activation and hyperactivation of neural areas not typically engaged during cognitive task performance compared with nondiabetic controls. Therefore, it appears that factors linked to insulin resistance play a complex role in cognitive function and cognitive decline.

Figure 2.

Cellular mechanism(s) by which proper peripheral insulin signaling through sedentary lifestyle (A) compared with exercise (B) may impact brain insulin signaling. AKT, serine/threonine protein kinase; ATP, adenosine triphosphate; BDNF, brain-derived neurotrophic factor; CD36, cluster of differentiation 36; CREB, cAMP response element-binding protein; FFA, free-fatty acids; GLUT1, Glucose transporter type 1; GLUT4, Glucose transporter type 4; GRB2, growth factor receptor bound protein 2; IKK, ikappaB kinase; IL-6, interleukin 6; IL-8, interleukin 8; IRS 1/2, insulin receptor substrate 1/insulin receptor substrate 2; JNK, c-Jun N-terminal kinase; MAPK, mitogen-activated protein kinases; mTOR, mammalian target of rapamycin; NADH, nicotinamide adenine dinucleotide hydrogen; NF-κB, Nuclear factor kappa-light-chain-enhancer of activated B cells; PGC-1α, proliferator-activated receptor-gamma coactivator-1alpha; PI3K, phosphoinositide 3-kinase; PKR, protein kinase R; Ras, guanosine-nucleotide-binding protein; ROS, reactive oxygen species; TNF-α, tumor necrosis factor α; TNFR, tumor necrosis factor receptor; TrKB, tropomyosin receptor kinase B. Created with BioRender with permission.

EXERCISE AND BRAIN INSULIN RESISTANCE

Aerobic and resistance exercise improve aerobic fitness and muscular strength and endurance. These changes in muscle function are clinically relevant because they are associated with decreased cognitive decline and brain degeneration (e.g., decreased gray matter volume) (73–78). In fact, slower walking speed is associated with hypoperfusion in adults with T2D and can predict cognitive impairment (58, 79). Interestingly, nonleisure time physical activity has also shown to be neuroprotective in adults, although this activity may need to be of moderate to high intensity compared with light intensity (80). This latter observation is noteworthy given interest in breaking up extended periods of sedentary behavior and identification of how movement, independent of exercise, can favorably impact health. Indeed, recent work in children and college-aged adults suggest that breaking up classroom sitting with physical activity may influence cognitive function and cerebral blood flow (81, 82). However, similar results were not observed in adults with excess body weight (83). In either case, aerobic or resistance movement improves muscle function through reduced systemic inflammation as well as favorable mitochondrial function and protein metabolism (84). However, few data exist specifically determining the effects of exercise (and/or physical activity) on brain insulin sensitivity. In rodent models, exercise has been reported to favorably impact insulin signaling in the brain (Fig. 2B). High-fat feeding in rodents is often used to induce insulin resistance and has been shown to decrease phosphorylation of insulin receptor-B, IRS-1, PI3-K, and AKT in the hippocampus (85). This decline in insulin signaling, however, was rescued with treadmill-based exercise, although not to comparable levels of exercise plus chow (carbohydrate-based) diet. Excitingly, exercise increased hippocampal neurogenesis during high-fat feeding as well. This points toward the ability of exercise to support insulin action on memory, learning, and tissue health, and to protect against obesity-related cognitive dysfunction. These findings parallel recent work in rodents showing that aerobic exercise increases brain insulin sensitivity, in part, through improved mitochondrial function (86). Gusdon et al. (87) further demonstrated that ∼3 wk of treadmill exercise training in 24-wk-old mice was associated with increased coupled complex I to complex III enzymatic activity in the brain. This increased enzymatic activity may improve brain mitochondrial function via the electron transport chain, ultimately leading to improvements in memory and cognitive function. Marosi et al. (88) reported that high-fat feeding elevated mitochondrial ROS that contributes to impaired insulin signaling in rodents. However, exercise was able to upregulate antioxidant pathways to protect against modifications of ROS on brain insulin sensitivity, a potential key deleterious factor in the aging hippocampus (88). Similarly, in a rodent model of AD where memory impairments exist, exercise was able to raise insulin signaling (e.g., IRS-1, AKT, and glycogen synthase 3-α) as well as reduce β-amyloid in conjunction with improved cognitive function (85). Improved insulin signaling may also relate to the preservation or deterioration of brain structure and function. For instance, Dietrich et al. (89) demonstrated in mice that exercise induces uncoupling protein 2 (UCP2) mRNA expression and mitochondrial oxygen consumption in the hippocampus. UCP2 is a mitochondrial protein that uncouples substrate oxidation from ATP synthesis. These changes in mitochondrial metabolism supported the increased bioenergetic adaptation to neurogenesis and synaptogenesis in response to exercise. Although these studies highlight the utility of exercise to positively impact brain insulin signaling, additional work in humans is warranted to understand how gains in brain insulin sensitivity relate to mitochondrial function as well as memory and executive function processes.

Exercise increases skeletal muscle (90), liver (91), and adipose tissue (92) insulin sensitivity to favorably regulate systemic glucose homeostasis in humans. Although few studies have focused specifically on the brain in humans, Kullmann et al. (93) recently showed that lifestyle interventions, consisting of exercise and diet counseling for 24 mo, increased brain insulin sensitivity following intranasal insulin administration in middle-aged people with prediabetes. Moreover, Honkala et al. (94) demonstrated that 2 wk of sprint interval training (SIT) in insulin-resistant, middle-aged adults decreased insulin-stimulated glucose uptake in cortical gray matter and all brain regions except the occipital lobe. Interestingly, these changes in brain insulin-stimulated glucose metabolism were not seen in adults who underwent moderate-intensity continuous training (MICT). This intensity-based effect on the brain is interesting since both SIT and MICT raised whole body insulin sensitivity via the euglycemic-hyperinsulinemic clamp. Despite this latter study suggesting exercise dose may impact tissue-specific insulin action on glucose metabolism, limited data exist to understand or support optimal mode, intensity, or timing of exercise on brain insulin sensitivity in people at risk for or with metabolic or brain disease.

EXERCISE AND COGNITION

The notion that various lifestyle behaviors such as exercise, diet, and social engagement impact cognitive functioning has continued to garner scientific attention (95). Early cross-sectional studies indicated that cognitive slowing associated with age were offset by greater amounts of leisure-time exercise and increased cardiorespiratory fitness (96, 97). Although these seminal studies were unable to establish a causal relationship between exercise and cognitive function, they provided foundational evidence to support the influence of physical activity and exercise for promoting cognitive function.

Kramer et al. (98) conducted one of the initial randomized controlled behavioral intervention trials of aerobic exercise (walking) on cognitive function among 124 previously sedentary 60- to 75-yr-old adults. Older adults assigned to aerobic exercise showed substantial improvements in performance on executive function tasks at 6 mo relative to those assigned to a stretching and toning comparator group. Interestingly, there was also a 5.1% increase in cardiorespiratory fitness over the course of the 6-mo intervention among adults in the walking condition compared with a 2.8% decrease in maximal aerobic capacity among the stretching and toning group. Their findings also pointed to the potential for task specificity in the effects of exercise on cognition, such that exercise resulted in larger effects for cognitive tasks involving executive function. Executive functions refer to higher-order cognitive functions that are supported by prefrontal and parietal brain regions. Subsequent meta-analyses (99, 100) confirmed a beneficial influence of exercise on executive functioning. Furthermore, combined aerobic and resistance exercise programs tend to result in larger effects than aerobic exercise performed alone, highlighting the potential cognitive benefit of alternative modes of exercise (e.g., resistance exercise). Interestingly, although adults with excess weight (101) and/or T2D (102) have demonstrated mild-to-moderate decrements in executive functioning, it may be possible to mitigate or reverse these deficits with exercise. Exercise has also been shown to influence cognitive domains of short- and long-term memory (103) and this may be one mechanism through which exercise decreases risk for age-related dementia and AD (104). Based on the collective evidence to date, the 2018 US Physical Activity Guidelines Committee concluded that sufficient evidence exists to indicate exercise favorably influences cognitive function in individuals across the lifespan, including those with various comorbid health conditions (105). Unfortunately, there are no clear public health guidelines for prescribing exercise to optimize its cognitive enhancing effects (106) and future research is warranted to examine the mode and dose of exercise to improve cognitive function, particularly among those at increased risk for accelerated cognitive decline.

INSULIN, EXERCISE, AND BRAIN PLASTICITY

Brain plasticity is the ability of the brain to undergo structural and functional changes in response to learning, life experience, and memory formation. This phenomenon has been studied in relation to sensory stimuli, diet, pharmaceuticals, and gonadal hormones. Still, few studies have examined the role of other hormones, such as insulin, on brain plasticity. The strength of signaling at the level of the synapse has been identified as a key underlying mechanism to facilitate changes in brain structure and function. Thus, understanding the effect of insulin at the synapse and on neurotransmitters may give insight into brain plasticity. Lee et al. (107) tested this mechanism in rodent models and found that insulin increases basal neurotransmitter release from the presynaptic terminals through activation of the PI3K/AKT/mTOR and Rac1 signaling pathways and promotes dendritic spine formations in the hippocampus. As previously mentioned, the hippocampus supports memory encoding and retrieval processes as well as various forms of learning. However, systemic hyperinsulinemia has been linked to decreased spatial memory and learning deficits in insulin-resistant, obese Zucker rats which further supports that too much insulin in the brain may have negative consequences (108). However, it is important to acknowledge that some evidence suggests that systematic hyperinsulinemia may contribute to decreased abundance and transport of insulin into the brain by downregulating the number of insulin receptors, thereby creating brain hypoinsulinemia (23). These latter findings highlight a need for additional work examining how insulin enters the brain across populations. Interestingly, Kamal et al. (108) reported that presynaptic structure and function were not the cause of impairment under hyperinsulinemic conditions in obese Zucker rats, but that diminished long-term potentiation (persistent strengthening of synapses) and decreased synaptic plasticity were associated with hyperinsulinemia.

Exercise, particularly aerobic exercise, improves brain structure and function across age groups and health states (109, 110). In one of the first studies to examine the influence of cardiorespiratory fitness on brain structure, Colcombe et al. (111) examined high-resolution magnetic resonance imaging (MRI) among 55 older adults and assessed differences in gray and white matter tissue density as a function of age and fitness. Consistent with the aging and brain structure literature, they reported age-related declines in the frontal, parietal, and temporal cortices. However, increased cardiorespiratory fitness had a sparing effect on gray matter in prefrontal, superior parietal, and temporal cortices. These brain regions are clinically relevant as they subserve aspects of higher-order cognition and executive functioning, such as working memory, cognitive flexibility, and inhibitory control. Colcombe et al. (75) later assigned sedentary older adults to either a moderate-intensity aerobic exercise group or a nonaerobic stretching and toning control group for 6 mo. Participants in both groups attended three 60-min sessions per week across the 6-mo intervention period. Using high spatial resolution MRI, increased gray matter volume was observed in prefrontal, parietal, and lateral temporal brain regions and increased white matter volume was found in the genu of the corpus callosum among the aerobic exercise group. These structural MRI data highlight the potential for aerobic exercise to attenuate the normal trajectory of age-related loss in brain tissue and enhance brain plasticity, particularly in regions critical for higher order cognition.

Erickson et al. (112) investigated whether individuals with higher levels of aerobic fitness displayed greater volume of the hippocampus and better spatial memory performance relative to their lesser fit counterparts. Their findings revealed that higher fitness levels were associated with larger left and right hippocampi after controlling for age, sex, and years of education among 165 nondemented older adults. In addition, larger hippocampi and higher fitness levels were correlated with better spatial memory performance. In a subsequent randomized control trial (77), Erickson et al. assigned 120 sedentary older adults without dementia to a moderate-intensity aerobic exercise intervention or a stretching and toning control group for one year. The aerobic exercise intervention was effective at offsetting the normal age-related deterioration in the hippocampus. In fact, there was a 2% increase in hippocampal volume following the one-year exercise program relative to a 1.4% decline in the control group. Increased hippocampal volume was associated with increased serum levels of brain-derived neurotrophic factor (BDNF) and improved memory performance. These neurogenic effects of exercise on the hippocampus have been supported by two meta-analyses of exercise on hippocampal volume. Firth et al. (113) reviewed 14 available studies and found a significant positive effect of exercise on left hippocampal volume in comparison to control conditions [g = 0.265, 95% confidence interval (CI) = 0.1–0.44]. A more recent meta-analysis demonstrated a small but significant effect of exercise on total hippocampal volume, with an effect size of g = 0.13 (114). This overall effect was driven by a significant decrease in hippocampal volume (−0.72%) in control or comparator groups, whereas an increase in hippocampal volume (1.2%) was observed across exercise treatment groups. The findings from both meta-analyses indicate that most randomized controlled trials to date have been performed on individuals over 65 years of age, suggesting that additional evidence is required before definitive statements can be made regarding the effect of exercise on the hippocampus across the lifespan or in individuals with insulin resistance, obesity, or T2D. Interestingly, individuals with overweight and obesity exhibit widespread alterations in the structure and function of the prefrontal cortex (115, 116) and hippocampus (117, 118). Behavioral weight loss interventions, including those that incorporate exercise, have been shown to improve cognitive functioning (119), highlighting that accelerated cognitive aging associated with obesity and T2D in mid-life may be reversible.

EXERCISE AND GROWTH FACTORS INFLUENCING THE BRAIN

Studies using rodent models have been instrumental in providing insight into the mechanisms through which exercise influences brain function. It is important to note that many of the neurobiological mechanisms that have been studied in animal models have been restricted to the hippocampus, although more recent work highlights influences across the brain and multiple organ systems (120). In the mid to late 1990s, evidence began to emerge indicating that the adult brain was capable of neuroplasticity and that new neurons were capable of being produced in the dentate gyrus of the hippocampus. Although the precise role of neural progenitor cells has long been debated (119), exercise emerged as an early and promising behavioral manipulation that resulted in an increased number and proliferation of new neurons in the hippocampus. Exercise may even be the critical component of the early environmental enrichment studies on neurogenesis (121). In addition to neurogenesis, exercise increases the number of connections or synapses between neurons (synaptogenesis) as well as blood supply through the proliferation of new vasculature (angiogenesis) in the cerebellum, hippocampus, motor cortex, frontal cortex, and basal ganglia (17, 122). Exercise also improves energy metabolism and neuroimmune modulation in the brain, and these neurotrophic processes may be mediated by increased production of principal growth factors (110, 122). Some of the most widely studied growth factors known to mediate the effects of exercise on brain health include BDNF, insulin-like growth factor (IGF-1), and vascular endothelial growth factor (VEGF). However, because of the widespread effects of exercise on whole body physiology, including all of the major organ systems (e.g., neuromuscular, skeletal, endocrine, cardiovascular, and respiratory), there are likely a host of different biologically plausible pathways through which exercise influences the brain and cognition (120).

BDNF is a neurotrophic factor that is considered essential for hippocampal function, synaptic plasticity, growth and differentiation of neurons, neuroprotection, as well as learning and memory. In a classic study, Neeper et al. (123) showed that physical exercise affects BDNF mRNA production not necessarily in brain regions involved in movement per se, but rather in the hippocampus and caudal cortex of rats. This is clinically relevant as these brain regions are essential to higher-level cognitive functions. Subsequent studies have confirmed that exercise increases BDNF levels in the hippocampus, cerebellum, and frontal cortex of mice, rats, and humans (124–126). Induction of BDNF in the hippocampus following exercise is regulated by other neuroendocrine systems, growth factors such as IGF-1, and neurotransmitter systems. Furthermore, BDNF signaling (by binding to its receptor) is a crucial mechanism underlying improved hippocampal-dependent learning in response to exercise (127) and blocking BDNF in animals by infusing an antibody that blocks BDNF activation of its receptor eliminates the benefit of exercise on learning (128). Inhibiting BDNF action also blocks the influence of exercise on downstream neurobiological pathways important for synaptic plasticity and cell proliferation (122). Collectively, these findings highlight a central role of BDNF in the effects of exercise on hippocampal plasticity, function, and learning.

In addition to BDNF, exercise also influences IGF-1 gene expression and protein levels, both in the periphery and in several brain regions (129). IGF-1 is a growth factor that influences neurogenic processes and reduced IGF-1 levels may contribute to age-associated cognitive decline (130). In addition, circulating IGF-1 levels are associated with hippocampal volume and performance on verbal learning and memory tasks in older adults (131). Circulating levels of IGF-1 are increased following exercise (132), can cross the BBB, and result in increased IGF-1 levels in the brain, particularly in the hippocampus (133). If circulating levels of IGF-1 are prevented from entering the central nervous system, many of the benefits from exercise, including neurogenesis, angiogenesis, and improved memory do not occur (127, 133). Together, the existing evidence suggests IGF-1 is another neurobiological mediator of the salutary effects of exercise on learning and memory.

VEGF is a potent angiogenic factor that is produced by skeletal muscles in response to exercise. Similar to IGF-1, circulating levels of VEGF cross the BBB into the central nervous system and contribute to exercise-dependent stimulation of neurogenesis and angiogenesis (127), although not all studies in humans support systemic effects of VEGF (134–136). In either case, studies in mice using antibodies that block entry of peripheral VEGF into the brain have shown that VEGF may be an important mediator of exercise-induced hippocampal neurogenesis (137).

The molecular mechanisms that underlie the beneficial effects of exercise remain poorly understood and recent work has suggested the influence of “exerkines,” a broad variety of signaling moieties released in response to acute and/or chronic exercise that exert their effects across multiple organ systems through endocrine, paracrine, and/or autocrine pathways (120). Consistent with the mechanisms discussed herein, it is possible that exercise results in complex yet critical interactions between insulin, IGF, VEGF, and BDNF on brain and body systems. Recent studies have begun to shed light on these potential interactions (138), and interestingly, have framed these purported interactions within the context of inflammation, stress, and resilience. Related to neurocognitive mechanisms of acute and chronic exercise, it is also important to examine regional versus global effects of these “exerkines” on the brain. Much of the research evidence of IGF, VEGF, and BDNF has focused on the hippocampus while exercise-related changes in insulin seem to have a broader cortical effect. Such findings may advance exercise therapy from a precision medicine approach.

FUTURE DIRECTION AND CONCLUSIONS

Exercise has emerged as a promising behavioral approach for positively influencing cognitive function and reducing the risk of age-related cognitive decline (Fig. 1). Aspects of cognitive function that may be improved include processing speed, attention, executive function, memory, and academic performance among youth. In addition, exercise reduces the risk of age-related neurodegeneration, dementia, and AD (104). Based on the available scientific evidence, experts have recently concluded that “exercise unequivocally influences the brain” (110). The research evidence from humans on exercise-dependent influences on cognition has largely focused on frontal-brain-dependent tasks (executive function and cognitive control), whereas animal studies have primarily assessed the effects of exercise on hippocampal-dependent learning and plasticity (122). Although unequivocal evidence supports the influence of exercise on the prefrontal cortex and hippocampus, additional evidence exists to support the effects of exercise on numerous structural and functional aspects of the brain. How exercise influences the brain is an area that warrants further attention though, as we present the hypothesis that insulin resistance is a key factor involved in cognitive improvement. In fact, several studies support that excess nutrients and energy surplus can lead to elevations in inflammation and oxidative stress generated by mitochondria that decrease brain insulin signaling. Thus, increased energy metabolism and decreased oxidative stress that accompany exercise represent fundamental mechanisms through which exercise may promote brain insulin signaling for cognitive gain. Interestingly, the role of brain insulin action is further evidenced by pharmaceutical intervention with metformin as insulin-sensitizing agent showing in some, but not all studies, favorable effects on cognitive function that may be modified by age, sex, and glycemia status (139–142). Furthermore, the sodium glucose transporter 2 inhibitors have gained recent attention for their glucose lowering benefit as well as cognitive function preservation effects in people with T2D (143). At the same time, the insulin secretagogue class of drugs referred to as GLP-1 agonists have also been shown to reduce the risk of developing AD in people with diabetes (144). However, caution should be used when considering the influence of these agents for promoting healthy lifestyles. It is also worth noting that a combination of exercise with diet and mental health activities should be considered as the FINGER study (145) demonstrated that cognitive and social activities as well as cardiometabolic risk were improved with this intensive lifestyle modification in middle-aged to older adults. Future prospective studies designed to tease out exercise, diet, and medication are warranted to understand the underlying issues and mechanisms related to cognitive health and brain insulin resistance. In turn, better understanding how insulin action on the brain affects cognitive health may promote new therapeutic approaches to combat obesity, T2D, CVD, and neurodegenerative diseases.

GRANTS

This work was supported by, in part, by the National Institutes of Health Grant RO1-HL130296 (to S. K. Malin).

DISCLOSURES

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

AUTHOR CONTRIBUTIONS

S.K.M. and N.R.S. conceived and designed research; S.K.M. and N.R.S. prepared figures; S.K.M., N.R.S., A.A.U., and B.L.A. drafted manuscript; S.K.M., N.R.S., A.A.U., and B.L.A. edited and revised manuscript; S.K.M., N.R.S., A.A.U., and B.L.A. approved final version of manuscript.