Obesity, leptin, and Alzheimer's disease

Abstract

Introduction

Understanding how obesity adversely affects the aging brain is hindered by the number of metabolic and hormonal pathways that are altered due to both obesity and aging. These changes include alterations in energy expenditure, reduced respiratory quotient, hyperlipidemia, hyperinsulinemia, glucose intolerance, low-grade inflammation, and changes in adipokine levels. Dissecting the relative role of each of these factors remains a challenge. Of the various changes associated with obesity, altered adipokine signaling may be one mechanism whereby obesity affects brain health. Adipokines are known to affect brain physiology and function, and both population and experimental studies suggest that changes in adipokine function may mitigate the pathogenesis of Alzheimer's disease (AD). The interaction between obesity and AD has been explored in several reviews, usually with an emphasis on insulin signaling pathways.^1–4 This review will emphasize the potential role of adipose tissue and adipokines in the pathogenesis of AD with a particular focus on leptin.

Neuropathology of Alzheimer's disease

AD is characterized by an insidious loss of memory and cognition leading to death within 10 years.^5–8 AD brains show neuronal loss and reactive gliosis involving limbic regions (including amygdala and hippocampus) and cerebral cortex with relative sparing of the basal ganglia, cerebellum, brainstem, and spinal cord.⁹ Affected brain regions show an abundance of intracellular and extracellular aggregates (Fig. 1). Intraneuronal aggregates called neurofibrillary tangles consist of hyperphosphorylated tau protein in the form of insoluble paired helical filaments.¹⁰ Extracellular deposits called amyloid plaques consist of Aβ peptides in the form of insoluble amyloid fibrils. Many amyloid plaques also contain swollen dystrophic tau-positive neurites and are thus called neuritic plaques. Both plaques and tangles stain avidly using dyes, such as thioflavin S or congo red, that bind to β-pleated sheet structures. Widespread involvement and a high density of plaques and tangles are diagnostic of AD.

Open in a separate window

Figure 1

Neuropathology of AD. Images from hematoxylin- and eosin-stained histologic sections (400× magnification) from affected human AD brain show: (A) granulovacuolar degeneration (arrowheads), (B) intracellular neurofibrillary tangles, and (C) waxy extracellular amyloid plaques. As neurofibrillary tangles and amyloid plaques are relatively subtle using routine stains, histologic diagnosis of AD is aided by using dyes that bind to amyloid structures or with antibodies specific for tau or Aβ(D) Thioflavin S staining (fluorescent green color, 100× magnification) of human AD necortex shows abundant extracellular amyloid plaques (arrows) and intraneuronal neurofibrillary tangles (arrowheads). Higher-power images (400× magnification) show (E) a neurofibrillary tangle and (F) an amyloid plaque. (G) Neurofibrillary tangles (arrows, 100×) are stained with antibodies that recognize phosphorylated tau protein epitopes. Note also the presence of clusters of phospho-tau immunoreactivity (circles) that label dystrophic neurites in association with amyloid plaques. (H) Amyloid plaques form in transgenic mouse overexpressing mutant APP that can be labeled with antibodies against Aβ (400× magnification).

Aβ peptides are generated by sequential proteolysis of the amyloid precursor protein (APP) by a series of endoproteolytic proteases historically called “secretases” prior to their cloning.⁹ β-secretase, now known as β-site APP cleaving enzyme (BACE), cleaves at the N-terminus of the Aβ sequence. An alternative cleavage pathway involving α-secretase results in proteolysis within the Aβ domain, precluding the generation of Aβ peptide. After α-or β-cleavage, the remaining C-terminal APP fragments are then cleaved by the γ-secretase enzymatic complex that consists of four proteins (presenilin, Aph1, Pen2, and nicastrin). Importantly, several mutations within the APP or presenilin genes result in autosomal dominant cerebral Aβ amyloidosis. These autosomal dominant amyloidoses usually result in clinical dementia and AD-like neuropathology, although other clinical syndromes and pathologies are observed, such as cerebral amyloid angiopathy, leading to recurrent hemorrhagic strokes. Finally, the vast majority of these mutations result in increased amyloidogenic Aβ peptide generation. Thus, Aβ is central to the pathogenesis of AD dementia.

Tau protein normally functions to bind and stabilize microtubules.¹⁰ Tau is prone to aggregation intracellularly where the protein is ubiquitinated and hyperphosphporylated at multiple sites. Although tau mutations have not been linked to AD dementia, several tau mutations are causative for other related neurodegenerative dementias indicating that tau dysfunction alone is sufficient for neuronal degeneration. Understanding the processes that influence amyloid plaque or neurofibrillary tangle formation may prove invaluable in mitigating AD and related neurodegenerative diseases.

Recent large-scale studies have been examining the relationship between clinical symptomology, various biomarkers of AD dementia, and brain pathology (Fig. 2).^5–8,11 Biomarkers currently under investigation include cerebrospinal fluid (CSF) Aβ, CSF tau, structural and functional brain imaging using MRI or PET, and synthetic amyloid imaging compounds that detect cerebral amyloid. The clinical spectrum of disease has been expanded in recognition of the fact that AD is likely a protracted and insidious disease. A predementia stage called “mild cognitive impairment” (MCI) can be defined using careful psychometric testing, sometimes coincident with a subjective decline in memory, reasoning, or visual perception that does not reach the diagnostic threshold of dementia.⁷ Individuals with MCI are at risk for subsequent dementia. However, even before clinical signs can be detected, abnormal biomarker values can be detected in cognitively normal individuals, leading to the concept of preclinical disease characterized by early biological changes including the onset of CNS pathology.⁸

Open in a separate window

Figure 2

Model of the clinical and biomarker progression of AD. Current efforts are underway to define the relationship between progression of AD (x-axis), various AD biomarkers (left y-axis), and clinical parameters associated with AD (right y-axis). Abnormal biomarker values related to the Aβ peptide (CSF Aβ measurements and/or amyloid dye imaging techniques, black line) indicate that abnormalities related to Aβ can be detected in presymptomatic individuals. Consequently, tau abnormalities and pathology may then develop (blue dotted line), followed by changes in brain structure and function (MRI and PET scans, purple dashed line). The onset of memory decline (empty green dotted line) heralds the clinical designation of “mild cognitive impairment” (MCI), which is typically followed by an overt decline in clinical function (empty red dashed line) and dementia, most commonly diagnosed as AD. Also reflected in this model is the association between obesity and dementia that is thought to occur in the presymptomatic phase of the disease as well as the association between weight loss (empty light blue line) and dementia that typically occurs close to and during the later stages of AD. Adapted and modified from work by Jack et al.¹¹

Body weight, obesity, and dementia

With the aging of the Baby Boomers, the number of Americans 65 and older will more than double from 2000 to 2050, and those 85 and older will increase by fivefold.^12–14 Advanced age is the strongest risk factor for AD, and AD is the most common aging-related neurodegenerative disease, with a projected 13 million or more afflicted by 2050 and with a cost of over $1 trillion unless effective interventions are implemented soon.^15–18 Obesity rates have also dramatically increased in the last 25 years such that greater than 1 in 3 adults are obese.¹⁹ Of all the demographic age groups, Baby Boomers have the highest rates of obesity at 40%.¹⁹ These demographics raise the public health concern of increases in aging-and obesity-related diseases.

The relationship between body weight and dementia is complex in that body weight has an age-dependent relationship with dementia (Fig. 2 and Table 1). Individuals with AD have lower body weight,²⁰ and it is plausible that dementia leads to a negative energy balance secondary to malnutrition. However, reductions in body weight is a harbinger of AD even before clinical symptoms of dementia are detected,^21,22 suggesting that loss of body weight may be a manifestation of early brain dysfunction. Furthermore, lower body mass indexes (BMIs) are associated with abnormal CSF Aβ and tau levels²³ and with increased CNS pathology at autopsy including neurofibrillary tangles and amyloid plaques.²⁴ Thus, weight loss is a consistent feature of AD dementia and correlates with the presence of abnormal biomarkers and increased brain pathology. However, weight loss associated with AD should be interpreted with caution since dual-energy x-ray absorptiometry studies suggest that weight loss is primarily due to sarcopenia and not loss of fat.²⁵ Thus total body weight and BMI may not be valid surrogate measures of obesity in the elderly. Given this propensity for weight loss, studies of elderly cohorts examining the relationship between obesity and AD are mixed,^{21,22,26–32} perhaps reflecting the difficulties in defining obesity in elderly cohorts based on anthropomorphic measurements such as BMI. For example, using the waist-to-hip ratio as a measure of central obesity reveals that obesity is associated with a significant hazards ratio of 2.5 for the development of AD while BMI in the same elderly cohort was not significantly associated with an increased risk for AD.³¹

Analysis of multiple biomarkers of AD has yielded a model in which AD is a protracted and insidious disease, with pathogenic processes occurring years if not decades prior to clinical symptoms.¹¹ Keeping this temporal distinction in mind, it is intriguing that mid-life obesity has been found to increase the risk of developing late-life AD (hazards or odds ratios ranging from ~ 1.4 to 3.6; see Table 1).^{27,30,33–37} These studies define obesity with different metrics, including BMI, waist circumference, sagittal abdominal diameter, and waist-to-hip ratios. Diabetes mellitus and vascular disease are also associated with dementia.¹ However, in as much as such factors can be treated as separate variables, the risk due to diabetes mellitus and vascular disease appears to be independent of the risk due to obesity. One important caveat to these studies is that most cohorts are clinically defined as having AD dementia without autopsy confirmation. Given the ubiquity of multiple brain pathologies in the elderly including a high prevalence of vascular disease, validation of these findings in autopsy-confirmed cohorts is desirable to define the relative contribution of vascular versus nonvascular pathology.^38,39 Nonetheless, the fact that obesity confers increased risk for AD during the preclinical stage of AD suggests that obesity may modulate biological pathways early in the pathogenesis of AD. Identifying factors that influence brain pathology before the onset of overt neurodegeneration provides an avenue for possible preventative intervention.

Diet and obesity in experimental AD models

Multiple AD mouse models have been generated, most of which overexpress mutant forms of APP and/or tau.⁴⁰ In general, expressing APP harboring mutations linked to human disease results in age-dependent accumulation of amyloid plaques and deficits in learning and memory behaviors (Fig. 1H).^41,42 Mutant tau transgenic models develop intracellular tau aggregates that can lead to age-dependent neuronal degeneration.⁴³ Although these models overexpress high levels of mutant protein and are therefore not entirely physiologic, these mice have proven invaluable in understanding the factors that influence the neuropathology of AD. APP transgenic mice have been fed various diets to determine whether diet can influence cerebral amyloid deposition. Several studies have documented increases in body weight or adiposity in APP transgenic mice in response to high-calorie diets. Diets range from high-fat diets to high-fat and high-cholesterol diets to high-sucrose diets. Remarkably, despite differences in dietary nutrient content, diet-induced obesity is consistently associated with an increase in cerebral amyloid pathology.^44–50 The only exception is a single study in which a Western diet was administered for only four weeks to examine relatively acute changes at an age (one to two months) before amyloid plaque pathology is found.⁵¹ Likewise, several dietary regimens associated with weight loss have been tested, including ketogenic diets and caloric-restriction diets. Again, studies that document weight loss uniformly show a correlation between weight reduction and decreased cerebral amyloid pathology.^52–56 These studies, performed on multiple different APP transgenic strains using a variety of diets from several independent laboratories, provide strong evidence that diet, body weight, and obesity modulate cerebral amyloid pathology.

Leptin and AD dementia

There are many possible factors and mechanisms that may account for this modulation of AD pathology. Indeed, several studies suggest that adipokines, and in particular leptin, may influence the pathogenesis of AD. A prospective study of 785 participants has indicated that higher circulating levels of leptin are associated with a reduction in AD incidence.⁵⁷ Higher leptin was associated with a lower incidence of dementia including AD dementia (hazard ratio for AD ~ 0.6), even when correcting for the waist-to-hip~ratio, BMI, and vascular risk factors in multivariate statistical models. Higher leptin levels were also associated with higher total cerebral brain volume in the subset of participants who underwent MRI imaging. Since leptin levels are known to fluctuate over time, it is remarkable that a significant association between leptin and AD could be detected using a single leptin measurement. The major determinant of circulating leptin levels is adipose tissue mass, and typically hyperleptinemia is associated with obesity and central leptin resistance. Thus, it is interesting that there was no association between leptin and AD incidence in the top quartile of participants based on waist-to-hip measurements or in individuals with a BMI > 30. Finally, the mean follow-up time of 8.3 years suggests that many leptin measurements were performed in the preclinical or MCI stages of the disease. This study suggests that leptin during the presymptomatic phase of AD, at least in nonobese individuals, may be neuroprotective and mitigate the progression of AD.

Secreted by adipose tissue, leptin regulates body weight by modulating LR-expressing neurons in the CNS, particularly within the hypothalamus and brainstem.^58–60 However, leptin has pleiotropic metabolic effects, regulating energy expenditure, feeding behavior, locomotor activity, bone mass, growth, thermogenesis, fertility, life span, adrenal function, and thyroid function. Thus, mice with congenital absence of leptin (ob/ob mice) or leptin's signaling receptor (db/db mice) exhibit a complex phenotype with abnormalities in virtually every organ system. These diverse effects are most congruous with the absence of leptin acting as a signal of starvation, triggering numerous compensatory changes that secondarily lead to obesity.⁶¹ Thus, leptin-deficient models are a complex hybrid of the starvation response and the numerous secondary effects of obesity.

Leptin acts through the longest isoform of leptin receptors (LRb), the only isoform containing the cytoplasmic signaling domain.^58–60 Leptin binding triggers phosphorylation of cytoplasmic tyrosine residues that initiate various signaling pathways including JAK2-STAT3, Erk1/2, and PI3K pathways (Fig. 3). Other signaling molecules may be regulated by leptin, such as AMP kinase (AMPK) and mammalian target of rapamycin complex 1 (mTOR1).^62–65 However, the signaling mechanisms by which leptin affects these molecules are not entirely known, and it is not known whether these pathways result from a direct effect of leptin on LRb.

Open in a separate window

Figure 3

Leptin receptor signaling. LRb is constitutively associated with Janus family of tyrosine kinase 2 (JAK2). Leptin binding to LRb triggers autophosporylation of JAK2^155–157 and subsequent phosphorylation of cytoplasmic LRb tyrosine residues, including Y1138 and Y985. Phosphorylation of Y1138 recruits STAT3, which is then phosphorylated and translocated into the nucleus to alter gene expression, including upregulation of suppressor of cytokine signaling 3 (SOCS3) expression.^142,155 SOCS3 binds to Y985 as a negative feedback regulator of LRb signaling.¹⁴¹ Y985 serves dual roles in that its phosphorylation promotes binding of the tyrosine phosphatase SHP-2, leading to downstream Erk1/2 phosphorylation/activation.^155,158 In cultured cells, LR activation also triggers phosphoinositide 3-kinase (PI3K)-Akt signaling,^86,89,93,94 although it is unclear whether Akt phosphorylation occurs in vivo.^98,159–161 Both Erk1/2 (via p90RSK) and Akt phosphorylation can lead to downstream GSK3 phsophorylation/inhibition.⁸⁶ Signaling molecules that are known to affect APP processing are highlighted in blue italics.

LRs are present in both hypothalamic and extrahypothalamic neurons, including neurons of the hippocampus and cerebral cortex.^66–74 The major metabolic effects are predominantly due to leptin action on hypothalamic and hindbrain neurons. However, several lines of evidence suggest that leptin has nonmetabolic CNS effects as well. For example, leptin does not exert any metabolic effects in mice prior to weaning despite a large postnatal surge, indicative of a function distinct from its role in metabolism.^75,76 The brains of ob/ob mice are smaller with reduced levels of synaptic proteins, abnormalities that are partially reversed by leptin treatment.⁷⁷ In normal elderly, circulating leptin levels correlate with gray matter volume in various brain regions including the hippocampus,^57,78 and inversely correlates with cognitive decline.⁷⁹ Leptin also reverses neurocognitive deficits and structural abnormalities in multiple brain regions in humans with congenital leptin deficiency.^80,81 In leptin-deficient individuals and people with recent weight loss (representing a state of relative leptin deficiency), exogenous leptin alters brain activation in response to food cues.^82–84 These findings indicate that leptin has strong effects on brain structure and function outside the hypothalamus.

A growing literature indicates that leptin is neurotrophic. Leptin promotes dendritic growth cones/filipodia outgrowth in hippocampal and cortical neurons in vitro^85,86 and exhibits trophic activity on neurons that regulate feeding behavior in vivo.^87,88 Leptin increases adult hippocampal neurogenesis⁸⁹ and enhances hippocampal long-term potentiation by enhancing NMDA receptor function in part through MAPK.⁹⁰ Indeed, db/db mice exhibit defects in hippocampal neuronal morphology and hippocampus-dependent learning and memory behaviors.^91,92 In addition to its neurotrophic role, leptin appears to be neuroprotective,⁷⁷ as seen in various models of neuronal injury including injury related to stroke, seizure, neurotrophin withdrawal, excitotoxicity, oxidative damage, apoptosis, 6-hyrdoxydopamine, and tumor necrosis factor-α.^93–99 The pleotropic effects of leptin on neuronal integrity and function makes it possible that leptin may have beneficial effects on the CNS independent of the pathogenic mechanisms of AD.

More relevant to the pathogenesis of AD, leptin reduces Aβ secretion in cultured neuronal cells or organotypic slices,^45,100,101 and chronic leptin treatment with supraphysiologic doses reduces Aβ levels in brain and serum of APP Tg mice.^45,102 The reduction of Aβ appears to reduce β-secretase expression and/or activity, and the reduction can be blocked by an inhibitor of the AMPK in cultured cells.^45,100,101 Leptin also reduces tau phosphorylation in cultured neuronal cells and organotypic cultures, and this effect was blocked by AMPK and glycogen synthase kinase-3 (GSK3) inhibitors.^{100,101,103,104} Several issues remain with regards to the mechanisms whereby leptin may modulate AD pathology. Although activating leptin pathways may influence Aβ or tau pathways, it remains unclear whether altered leptin signaling is physiologically relevant in terms of AD pathogenesis. Toward that end, crossing APP transgenic mice onto a leptin-deficient ob/ob background results in worsening of cognitive function, enhancement of cerebrovascular inflammation, and increased cerebral amyloid angiopathy.¹⁰⁵ A fundamental issue with understanding the mechanisms of leptin action in vivo is the vast number of metabolic and physiologic parameters that are regulated by leptin. Thus ob/ob mice represent an extreme phenotype compared to diet-induced obesity models. Another question that deserves attention reflects the inherent complexity of CNS circuitry, and thus it is unknown whether leptin acts directly on cortical or hippocampal neurons to inhibit amyloid plaque deposition as opposed to an indirect effect via hypothalamic relay neurons. Finally, most cultured cell lines do not express the long LRb isoform of the leptin receptor, and thus it is important to determine whether leptin's effects are mediated through LRb and the known downstream LRb signaling pathways. For example, although it is proposed that leptin modulates Aβ and tau through AMPK, it is unclear how leptin activates AMPK. In skeletal muscle, much of AMPK activation in vivo is through autonomic innervation and not a direct effect of leptin binding to LRb.⁶⁴ Leptin actually decreases AMPK activity in the hypothalamus and knockout of AMPK in POMC and AgRP expressing hypothalamic neurons has no effect on their leptin responsivity.^63,106

Many of the other receptor-signaling molecules downstream of LRb are known to affect secretase activity or APP trafficking (Fig. 3, highlighted in blue). First, STAT3 increases BACE expression by binding to its promoter and is also thought to mediate Aβ toxicity.^107–110 Second, Erk1/2 activation increases α-secretase activity, and may inhibit β- and γ-secretase activity.^111–120 PI3K activation increases APP trafficking and the secretion of APP metabolites,^121–124 and constitutively active Akt inhibits APP trafficking by feedback inhibition of PI3K.¹²⁵ Finally, inhibition of GSK3 results in an inhibition of γ-secretase.^126,127 Therefore, leptin signaling can potentially intersect with APP processing pathways at multiple levels, and it remains to be determined which pathway mediates leptin's antiamyloidogenic effects in vivo.

One additional consideration is the relationship between obesity, aging, and leptin. Circulating leptin levels are high in obese humans and rodents in correlation with adiposity.⁶⁰ Aging also results in increased leptin levels and adiposity.^128–130 The maintenance of increased fat despite hyperleptinemia indicates that obesity and aging are both states of leptin resistance. The exact mechanisms of leptin resistance are not completely understood but are in part due to defective leptin transport across the blood–brain barrier and in part are intrinsic to leptin-responsive neurons.^129–140 Intrinsic neuronal leptin resistance has been linked to increased feedback inhibition by SOCS3 and increased protein tyrosine phosphatase 1B (PTP1B) activity, both of which dampen downstream LR signaling pathways.^141–151 This interrelationship between peripheral leptin, central leptin resistance, obesity, and aging should be considered in future clinical and experimental studies on leptin and AD.

Conclusions

Advances in the fields of endocrinology and neuropathology are beginning to reveal a complex relationship between peripheral metabolic factors and brain health. Although obesity increases the risk for AD in humans and diet-induced obesity modulates AD pathology in transgenic mice, the mechanisms that account for these phenomena are not entirely known. In terms of leptin, increased peripheral leptin is associated with reduced incidence of AD in the nonobese elderly, and treating AD transgenic mice with leptin ameliorates AD pathology. It remains to be determined whether central leptin resistance associated with aging and obesity enhances amyloid plaque deposition. Also, the effects of other adipokines, in particular the inflammatory adipokines (e.g., adiponectin, cytokines, and complement), have received little attention with regards to their contribution to AD. Adiponectin is thought to be complementary to the actions of leptin and exhibits anti-inflammatory properties. However, only two small studies have been reported with conflicting results as to whether changes in plasma adiponectin are related to MCI or AD.^152,153 Thus, further studies of the relationship between adipokines and AD are warranted. Finally, despite the emphasis on leptin in this review, several genes associated with lipid homeostasis are thought to influence AD risk including genes encoding apolipoprotein E, apolipoprotein J (clusterin), and sortilin-related receptor.¹⁵⁴ Undoubtedly, multiple metabolic and hormonal changes associated with obesity and aging will prove to influence the brain in health and disease. Together, this burgeoning field is proving that the brain is under the influence of peripheral metabolism and that an integrated physiologic approach to understanding brain health and disease may reveal novel mechanisms that may be amenable to therapeutic intervention.

Acknowledgments

Footnotes

References